Sunteți pe pagina 1din 7

2005-01-2361

Investigations on Whistle Noise in Automotive


Exhaust System Mufflers
Rolf Jebasinski, Sascha Leng, Marco Jess and Thomas Rose
J. Eberspaecher GmbH & Co KG
Copyright 2005 SAE International
ABSTRACT
Automotive exhaust mufflers are not only attenuating
noise, but can, in fact, along with other exhaust
components, generate noise within the exhaust system.
This type of noise is commonly called flow generated
noise to distinguish it from the pulsation noise excited
by the combustion engine. Flow generated noise is
normally very broad band but it can also have tonal
regions. This tonal flow noise can sound like a whistle. In
this contribution an experimental study of the generation
and prevention of such whistle tones will be presented.
The study was focused on perforated pipes in mufflers
with different perforation patterns, hole sizes and hole
shapes. The noise spectra have been measured on a
flow bench at room temperature as well as under real
(hot) conditions on a vehicle. The results will be
correlated in terms of frequency and Strouhal number.
Concepts to avoid or suppress these tones will also be
discussed.
INTRODUCTION
Sound quality aspects are playing an important role in
today's passenger car development, and exhaust
systems have a significant impact on this sound quality.
Therefore, not only the level of an exhaust system but
also its sound quality is important. The exhaust system
contains two different sound characteristics: pulsation
noise (excited from the engine) and flow noise (excited
within exhaust components). Both contribute to the
overall noise level as well as to the sound quality.
As a first step of sound design a sound cleaning of the
annoying noise content is necessary. One example of
this annoyance is the whistle, a narrow bandwidth
sometimes even tonal aerodynamic sound, at
frequencies between 1-15 kHz. It is not only subjectively
annoying it can also dominate the overall tailpipe noise
level of an exhaust system.
The generation mechanism of whistle is well known and
based on the exhaust gas flow grazing either a
perforated area or a side resonator attached to a muffler
chamber, which leads to shear layer instabilities [1].
These instabilities can grow to a vortex sheet and couple
at the interface between pipe perforations and the
muffler chamber under discrete flow/geometry conditions
to acoustic resonances of the cavity, which in turn can
lead to high sound pressure amplitudes. In fact, the
muffler is, in such cases, a noise generator.
These configurations are usually grouped into shallow
cavity oscillations, which are resonances due to
longitudinal standing waves, and deep cavity
oscillations, which are resonances due to transverse
standing waves [1]. Cavities are shallow if the length (L)
of the branch opening in the pipe is larger than the depth
(D) of the attached cavity (L>D) and they are deep if
L<D. Since the hole size of perforations in muffler pipes
are very small compared to the transverse size of typical
muffler chambers, whistling of perforated pipes in
mufflers can be regarded as deep cavity oscillation.
Theoretical models exist for deep cavity resonators with
a single opening like Helmholtz [2-5] or quarter wave
resonators [6]. These models can predict the acoustic
and hydrodynamic conditions for resonance by means of
an upper and lower bound of the Strouhal number
St=f*d/v (f is the frequency of vortex shedding, d is the
characteristic length of the object in the fluid flow and v
is the velocity of the fluid). Strouhal numbers found by
these models and other experimental work [7-9] are in
the range of 0.25-0.5 for the first vortex mode and 0.6-
0.9 for the second vortex mode. However, it is unclear if
these models can be applied to perforated sections in a
muffler.
Another way of approaching the dilemma could be
numerical simulation with CFD programs. It has been
shown that problems like flow excited Helmholtz and
quarter wave resonators can be modeled and calculated
with CFD [10-12], but it is questionable if this method is
appropriate for standard development work. The time
needed for meshing every hole in a perforated section
as well as the run time of these transient CFD
calculations are quite high. In addition one has to
consider numerical dissipation and dispersion effects
which could deteriorate the result. Thus it is not easily
possible to predict whistle noise in the exhaust systems
and therefore it must be approached experimentally.
The objective of the experimental study was to classify
certain types of mufflers in terms of whistle noise
generation probability. Different types of experiments
have been conducted. In order to correlate cavity modes
of the mufflers with whistle generation transmission loss
and the tailpipe noise of the mufflers on a cold flow
bench have been measured. These experiments will be
compared to tailpipe noise measurements on a vehicle
with hot gas flow.
Besides a general understanding that perforated
sections in mufflers tend to whistle under certain
conditions, measures to avoid or suppress this whistling
are important in muffler development. Whistle noise can
be attenuated by absorption mufflers positioned
downstream from the whistle source. Unfortunately that
is not always possible due to package constraints or
customer requirements. Therefore other options to
suppress whistling will be investigated in this
contribution.
EXPERIMENTAL SET-UP
In order to correlate flow acoustics and cavity modes of
the mufflers measurements on a flow bench and on a
transmission loss bench at room temperature were
performed. Figure 1 provides a sketch of both test
benches.
200 mm
@ 45
Blower
Muffler
Microphone
200 700 kg/h
(a)
White Noise
Speaker
Anechoic
Termination
Microphones
Muffler
Pipe radius
24 mm 24 mm
Microphones (b)
200 mm
@ 45
Blower
Muffler
Microphone
200 700 kg/h
(a)
200 mm
@ 45
Blower
Muffler
Microphone
200 700 kg/h
(a)
White Noise
Speaker
Anechoic
Termination
Microphones
Muffler
Pipe radius
24 mm 24 mm
Microphones (b)
White Noise
Speaker
Anechoic
Termination
Microphones
Muffler
Pipe radius
24 mm 24 mm
Microphones (b)

Figure 1: Flow test bench (a) and Transmission Loss test bench (b)
As shown in Figure 1 (a) the tailpipe noise of each
muffler was measured at 200 mm from the tailpipe at 45
degree at room temperature (23C). The airflow
generated from the blower has been varied from 200
kg/h to 700 kg/h. Since the mufflers internal pipe
diameter is 52 mm the gas velocity varied between 23
m/s and 80 m/s. The maximum Mach number reached at
the flow bench is 0.23.
Transmission loss (TL) measurements were also
performed by using the 4 microphone method [13].
Special attention was required to expand the frequency
range of the TL measurements. As shown in Fig. 1 (b)
the microphones were mounted within the 52 mm
diameter pipe. With a microphone distance of 24 mm the
upper frequency limit is 7170 Hz for plane wave
propagation. Usually the microphones are mounted flush
with the pipe wall. With such a configuration we
observed very low coherence between the microphone
signals at about 3900 Hz and 6300 Hz. These
frequencies correspond to the first two diametral (or axis
symmetric) modes which can propagate in a 52 mm
diameter pipe. So it is likely that the propagation of these
two transverse modes deteriorate the TL signal.
Therefore we moved the microphones in the nodal line
of the first two axis symmetric modes, which is at the
middle line of the pipe. With this configuration the
coherence between the microphone signals was over
the whole measurement range (6400 Hz) better than
0.9.
To compare the results from the room temperature
experiments with actual conditions in a vehicle, tailpipe
noise measurements were performed on a chassis
dynamometer. A V6 2.8 Liter gasoline vehicle was used
for the measurements. The mass flow range of the
engine under full load acceleration was comparable to
the range investigated at the flow bench. The same
mufflers, that were used in the flow bench and TL
measurements, were used as a rear muffler in the
vehicle. Tailpipe noise was measured 500 mm from the
tailpipe orifice at a 45 angle over the speed range
(1000-6000 rev/min) with full load acceleration. In
addition to the tailpipe noise measurements, the static
pressure and the gas temperature in front of the muffler
were measured. The gas temperature range was 250-
350C independent of the muffler used. By using the
measured data the maximum Mach number is calculated
to be 0.3 which is somewhat higher than on the flow
bench.
L
M
D
M
L
perf
D
P
=
5
2

m
m
L
M
D
M
L
perf
D
P
=
5
2

m
m

Figure 2: Layout test muffler
For the experiments a 17 Liter and a 4 Liter muffler with
perforated inner pipe and a circular cross section
(Muffler 1 with D
M
=213 mm / L
M
=480 mm, Muffler 2 with
D
M
=148 mm / L
M
=240 mm) were used (see figure 2).
The pipe was centered in the middle of the muffler cross
section. The diameter (D
P
) of the inner pipe was in each
case 52 mm. The following experiments have been
performed:
Variation of perforation length L
perf

Variation of porosity and hole size with fixed
perforation length
Variation of hole shape (circular, slot, two types of
louvers) with fixed perforation length (See figure 3)
Slot
Circular hole
Fish-scale
Bridge
Pipe
L
S
W
S
< L
S
d
d
W
FS
W
B
L
FS
L
B
t
FS
t
B
(a)
(b)
(c) (d)
Slot
Circular hole
Fish-scale
Bridge
Pipe
L
S
W
S
< L
S
d
d
W
FS
W
B
L
FS
L
B
t
FS
t
B
(a)
(b)
(c) (d)

Figure 3: Hole shapes: (a) Slot L
S
xW
S
=19x2mm
2
(opening area
38mm
2
), (b) Louver type Fish-Scale L
FS
xW
FS
xt
FS
=8x8x5.5mm
3

(effective opening area 20.5 mm
2
), (c) circular hole d=3.5 mm (5mm
and 7 mm), (d) Louver type Bridge L
B
xW
B
xt
B
=19x5x2.5mm
3
(effective
opening area 38 mm
2
).
The perforation was in each case uniform over length
and perimeter.
RESULTS AND DISCUSSION
As already mentioned in the introduction, it is well know
that the mechanism of whistle noise can be attributed to
a locking-in of the vortex shedding of flow grazing over
the perforated area with acoustic resonances of the
attached cavity.
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
[m/s]
60
70
80
90
100
[dB(A)]
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
60
70
80
90
100
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
60
70
80
90
100
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
60
70
80
90
100
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
60
70
80
90
100
800 1600 2400 3200 4000 4800 5600 6400 [Hz]
30
40
50
60
70
80
60
70
80
90
100
L
perf
= 40 mm (=0.8*D
pipe
)
L
perf
= 50 mm (=1.0*D
pipe
)
L
perf
= 100 mm (=2.0*D
pipe
)
L
perf
= 150 mm (=2.5*D
pipe
)
L
perf
= 240 mm (=4.6*D
pipe
)
L
perf
= 480 mm (=9.2*D
pipe
)
(a)
(b)
(c)
(d)
(e)
(f)
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
[m/s]
60
70
80
90
100
[dB(A)]
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
60
70
80
90
100
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
60
70
80
90
100
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
60
70
80
90
100
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
60
70
80
90
100
800 1600 2400 3200 4000 4800 5600 6400 [Hz]
30
40
50
60
70
80
60
70
80
90
100
L
perf
= 40 mm (=0.8*D
pipe
)
L
perf
= 50 mm (=1.0*D
pipe
)
L
perf
= 100 mm (=2.0*D
pipe
)
L
perf
= 150 mm (=2.5*D
pipe
)
L
perf
= 240 mm (=4.6*D
pipe
)
L
perf
= 480 mm (=9.2*D
pipe
)
(a)
(b)
(c)
(d)
(e)
(f)

Figure 4: Tailpipe noise spectra from the flow bench with variation of
the perforation length L
perf
in Muffler 1 and constant porosity (15%)
and hole size (3.5 mm)
Since the source excitation is based on a developed
vortex sheet over the holes, experiments were
conducted to exploit the necessary perforation length for
whistle tone generation on the cold flow test bench. In
figure 4, sound pressure level spectra of Muffler 1 with
different perforation length starting from L
perf
/D
P
= 0.8 to
L
perf
/D
P
=9.2 at a constant porosity of 15 % (hole size 3.5
mm) are shown in dependency of the air mass flow. As
illustrated in Figure 4 (a), with L
perf
=0.8*D
P
there is no
visible whistle resonance, but as the perforated length
increases, the whistle resonances become more visible.
At L
perf
=1.0*D
P
(Figure 4 (b)) a visible light whistle
develops at about 4800 Hz at a mass flow of 600 kg/h.
With L
perf
=2.5*D
P
a second flow/frequency regime
resonance with high amplitudes develops around 3200
Hz at 400 kg/h. A third whistle region starts with
L
perf
=4.6*D
P
at 2000 Hz and 230 kg/h. The same general
behavior was observed for Muffler 1 with 30 % porosity
perforation and the smaller Muffler 2 with 15 % porosity
perforation.
As demonstrated in Figure 4 whistle can occur with
perforation length as short as the perforated pipe
diameter. With increased perforation length the
excitation is strong enough to stimulate other cavity
modes at lower mass flows.
The Strouhal number St
d
=f*d/v (v: mean flow velocity, d:
hole diameter, f: frequency) is the standard metric to
classify flow excited cavities in terms of the frequency
and gas velocity. Besides frequency and gas flow, the
size of the flow disturbing object, such as the hole size,
is determining the Strouhal number. To verify this
dependency, experiments were conducted by varying
the hole diameter and having the perforated length
constant. Muffler 1 was used for this experiment. The
results are illustrated in Figure 5.
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
[m/s]
60
70
80
90
100
[dB(A)]
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
60
70
80
90
100
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
60
70
80
90
100
800 1600 2400 3200 4000 4800 5600 6400 [Hz]
30
40
50
60
70
80
60
70
80
90
100
S
t
d
=
0
.2
-0
.3
3
S
t
d
=
0
.2
-0
.3
3
S
t d
=
0
.
2
-
0
.
3
3
S
t d
=
0
.
2
-
0
.
3
3
St
d
=0.45-0.66
St
d
=0.45-0.66
St
d
=0.45-0.66
St d
=0.45-0.66
8 % porosity
d=5.0 mm
8 % porosity
d=7.0 mm
15 % porosity
d=3.5 mm
8 % porosity
d=3.5 mm
(a)
(b)
(c)
(d)
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
[m/s]
60
70
80
90
100
[dB(A)]
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
60
70
80
90
100
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
60
70
80
90
100
800 1600 2400 3200 4000 4800 5600 6400 [Hz]
30
40
50
60
70
80
60
70
80
90
100
S
t
d
=
0
.2
-0
.3
3
S
t
d
=
0
.2
-0
.3
3
S
t d
=
0
.
2
-
0
.
3
3
S
t d
=
0
.
2
-
0
.
3
3
St
d
=0.45-0.66
St
d
=0.45-0.66
St
d
=0.45-0.66
St d
=0.45-0.66
8 % porosity
d=5.0 mm
8 % porosity
d=7.0 mm
15 % porosity
d=3.5 mm
8 % porosity
d=3.5 mm
(a)
(b)
(c)
(d)

Figure 5: Tail-pipe noise spectra from the flow bench with variation of
the hole size from 3.5 mm to 7 mm in Muffler 1 and constant
perforation length L
perf
=L
M

St
d
can be regarded as slope of a line in the sound
pressure spectras because the air mass flow is linearly
proportional to the flow velocity at the cold flow test
bench. By drawing an envelope line to the upper and
lower bound of the whistle tones the Strouhal Number
can be determined. As shown in Figure 5 (a) St
d
is 0.2
for the upper bound and 0.33 for the lower bound in
case of Muffler 1 with circular hole perforation of d=3.5
mm. Thus, the range for whistle tone generation lies
within St
d
=0.2-0.33 for this particular muffler. Decreasing
porosity to 8 % while maintaining the hole diameter to
3.5 mm does not change that St
d
range as illustrated in
Figure 5 (b), but the whistle tones are at other
frequencies and less pronounced.
Changing the hole size to 5 mm (see Figure 5 (c)) or
even 7 mm (see Figure 5 (d)) moves the whistle tone to
lower frequencies as expected. The Strouhal range
stays at St
d
=0.2-0.33 for the upper and lower bounds of
the whistle tones, which can be attributed to the first
vortex mode. Another excitation mode (second vortex
mode) at St
d
=0.45-0.66 can be also seen in Figure 5, but
the whistle tones in this region are very weak. The same
behavior was also observed for Muffler 2.
In most of the published papers [2-9] on deep cavities
the Strouhal numbers for the first and second vortex
mode are reported somewhat higher (0.3-0.5 and 0.6-
0.9) but St
d
=0.2-0.3 were also reported for perforated
pipes with circular holes [14-15] which is similar to the
St
d
demonstrated in figure 5.
Figure 6 shows the frequency spectra of Muffler 1 with
3.5 mm holes and 15 % porosity compared to the
transmission loss. It is obvious that the whistle tones
correspond to muffler resonances at 2000 Hz, 3000-
3200 Hz and a wide region of smaller resonances from
4600-5600 Hz. Not all resonances visible in the
transmission loss are excited in the first Strouhal band
0.2-0.33. The strong resonance at 3800 Hz, for example,
seems only to appear very weak in the second Strouhal
band 0.45-0.66. The cavity modes excited in each of the
above cases are transverse modes of the muffler. These
transverse modes strongly depend on the overall shape
of the muffler. For circular expansion chambers the
transverse modes can be calculated analytically [16].
The first three transverse modes for an expansion
chamber with diameter of Muffler 1 are the two diametral
modes (1,0) and (2,0) at 934 Hz and 1544 Hz,
respectively, and the radial mode (0,1) at 1946 Hz.
Therefore the first whistling at 2000 Hz can be attributed
to the first radial mode (0,1). In fact it is not very likely
that one of the diametral modes will be excited, since we
have a highly symmetric configuration with the pipe
entering the circular muffler in the middle of the cross
section. At higher frequencies, many other mode shapes
can exist and therefore it is hard to decide which mode
can be attributed to a certain resonance peak in the
transmission loss.
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
[m/s]
60
70
80
90
100
[dB(A)]
S
t
d
=
0
.2
-0
.3
3
St
d
=0.45-0.66
800 1600 2400 3200 4000 4800 5600 6400
0
10
20
30
40
50
T
L

[
d
B
]
Frequency [Hz]
(a)
(b)
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
[m/s]
60
70
80
90
100
[dB(A)]
S
t
d
=
0
.2
-0
.3
3
St
d
=0.45-0.66
800 1600 2400 3200 4000 4800 5600 6400
0
10
20
30
40
50
T
L

[
d
B
]
Frequency [Hz]
(a)
(b)
Figure 6: Comparison tail-pipe noise spectrum at the flow bench (a)
and transmission loss (b) of Muffler 1 with circular hole perforation
(d=3.5mm, 15 % porosity, L
perf
=L
M
)
From the foregoing investigations it is clear that circular
holes are an effective excitation source for whistle noise.
Another type of hole shape is the rectangular slot, which
was believed by many exhaust manufacturers as less
prone to whistling. The effect of slot perforation (see also
figure 7) with a porosity of 8 % in Muffler 1 on flow noise
and on the mufflers transmission loss is shown in figure
7. The frequencies of the whistle have changed
significantly. Strong whistling is apparent at 2550-3000
Hz where several resonances are visible in the
transmission loss. In addition the resonance at 1500 Hz
is excited quite strongly at about 370 kg/h and again at
690 kg/h. The Strouhal band in the strong whistling
regime is between St
L
=0.68-0.84 (Strouhal number now
based on the slot length L=19 mm). This seems to be
the second vortex mode since at 690 kg/h a whistling at
1500 Hz was also found, which could be attributed to the
first vortex mode at St
L
= 0.34-0.42. No other whistle
tones are noticeable in this Strouhal band, which
corresponds to the transmission loss where no
resonances are visible up to 1500 Hz. From the
difference in transmission loss from fig 6 and 7, it is
obvious that the perforation hole shape and possibly the
porosity have an influence on the excitation of
transverse modes in the muffler. So even with such a
simple muffler shape, it is difficult to calculate transverse
mode analytically. In order to predict the first modes
which can be excited to whistling, it will be necessary to
calculate them with Finite Elements or Boundary
Element programs taking into account the perforated
pipe.
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
[m/s]
60
70
80
90
100
[dB(A)]
S
t L
=
0
.6
8
-0
.8
4
800 1600 2400 3200 4000 4800 5600 6400
0
10
20
30
40
50
T
L

[
d
B
]
Frequency [Hz]
S
t L
=
0
.3
4
-
0
.4
6 (a)
(b)
800 1600 2400 3200 4000 4800 5600 6400
30
40
50
60
70
80
[m/s]
60
70
80
90
100
[dB(A)]
S
t L
=
0
.6
8
-0
.8
4
800 1600 2400 3200 4000 4800 5600 6400
0
10
20
30
40
50
T
L

[
d
B
]
Frequency [Hz]
S
t L
=
0
.3
4
-
0
.4
6 (a)
(b)
Figure 7: Comparison tail-pipe noise spectrum at the flow bench (a)
and transmission loss (b) of Muffler 1 with slot perforation (L
S
=19 mm,
8 % porosity, L
perf
=L
M
)
The upper and lower bound Strouhal number St for flow
over deep cavities can be calculated from the equations
below [5].
( )
( )
|
.
|

\
|
=
(

2
n 2
S sin S
S cos 1
tan S
1

where

=
2
S
v
u
St
c

with u
c
being the vortex convection velocity, v the mean
gas velocity and n the vortex mode. For the typical
vortex convection speed of u
c
=0.4*v [1, 6] the equations
result in St=0.28-0.48 for the first vortex mode and
St=0.68-0.88 for the second vortex mode which is quite
close to the experimentally observed values for the slot
perforation. The Strouhal numbers for the circular holes
are lower and the range for the first and second vortex
mode is broader compared to the slot perforation.
All experiments shown so far have not been performed
under real conditions, i.e. excited from the hot gas flow
of an engine. In figure 8, the sound pressure level
spectra of Muffler 2 at the cold flow bench and on a V6
gasoline
2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
200
300
400
500
600
700
[kg/h]
60
70
80
90
100
[dB(A)]
2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 [Hz]
1000
2000
3000
4000
5000
6000
[1/min]
60
70
80
90
100
79
68
57
45
34
23
139
120
88
62
36
15
S
t d
=
0
.
2
-
0
.
3
3
S
t
d
=0.2-0.33
St d
=0.45-0.66
St
d
=0.45-0.66
Gas temp.: 23C
Gas temp.: 250-350C
[m/s]
[m/s]
(a)
(b)
2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
200
300
400
500
600
700
[kg/h]
60
70
80
90
100
[dB(A)]
2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 [Hz]
1000
2000
3000
4000
5000
6000
[1/min]
60
70
80
90
100
79
68
57
45
34
23
139
120
88
62
36
15
S
t d
=
0
.
2
-
0
.
3
3
S
t
d
=0.2-0.33
St d
=0.45-0.66
St
d
=0.45-0.66
Gas temp.: 23C
Gas temp.: 250-350C
[m/s]
[m/s]
(a)
(b)

Figure 8: Comparison tail-pipe noise spectrum on the cold flow bench
(a) and on the roller dynamometer with V6 engine (b) of Muffler 2 with
Hole perforation (d=3.5 mm, 15 % porosity and L
perf
=L
M
). The mass
flow and speed axis was rescaled to a flow velocity scale (see blue y-
axis labels)
engine are compared. The gas temperature at the cold
flow bench was 23C, whereas it was 250-350C on the
V6 engine depending on the engine speed. For a better
comparison the mass flow and the engine speed are
converted to gas velocity in the perforated muffler pipe.
For the flow bench this is just a simple conversion
because the temperature is constant, whereas the gas
temperature of the V6 increases with speed. The
Strouhal bands from the engine measurements are not
linear, since the temperature change from 1000-6000
rpm is about 100C. A straight Strouhal band has been
drawn for simplicity in Figure 8 (b). Nevertheless, it can
be seen that the first Strouhal band is again at 0.2-0.33,
whereas the whistle tones are at different frequencies
compared to Figure 8 (a) with cold flow. Due to the
different transverse modes in the smaller Muffler 2, the
first whistle is at about 2500 Hz which is close to the
(0,1) radial mode at 2800 Hz for D
M
= 148 mm. With
higher gas temperature the whistle tones are shifting to
higher frequencies and gas velocities, but they remain in
the Strouhal band. In addition all whistle regions are
appearing again but at higher frequencies. For example
the whistle at 2500 Hz in case of cold flow moves with
hot gas to 3100 Hz. That frequency shift corresponds
well to the change in speed of sound at the different
temperatures from 344 m/s (at 23C) to 447 m/s (at
250C). A similar behavior was found for other
perforations pattern in Muffler 2.
The foregoing experiments have shown that circular
holes and even slots are a strong source for whistle
excitation in empty mufflers. If there is no absorption
muffler downstream from the whistling muffler, the
whistle will propagate to the tailpipe and radiate into the
atmosphere. Quite often it is not possible to use
absorbing material. Therefore measures to avoid
creating whistle would be quite useful. Reference [6 and
17] have tried different interface geometries of a
whistling side resonator while others [14] reported less
whistling noise with louver perforation. As a first try, little
dimples (3 on circumference of the pipe) have been
pressed in the pipe within a distance of D
P
(see sketch in
figure 9).
Rings with 3 dimples on circumference
t
D
Distance rings = D
P
Flow
D
P
Rings with 3 dimples on circumference
t
D
t
D
Distance rings = D
P
Flow
D
P
Figure 9: Dimple-rings in hole (d=3.5 mm) perforated pipe with dimple
depth t
D
= 5 mm and distance of the rings equal to pipe diameter
2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
1000
2000
3000
4000
5000
6000
[1/min]
60
70
80
90
100
[dB(A)]
2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
1000
2000
3000
4000
5000
6000
60
70
80
90
100
2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
1000
2000
3000
4000
5000
6000
60
70
80
90
100
2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 [Hz]
1000
2000
3000
4000
5000
6000
60
70
80
90
100
Hole shape fig. 3(c), circular holes d=3.5 mm
Hole shape fig. 3(c), circular holes d=3.5 mm wi th dimple rings as in fig. 9
Hole shape fig. 3(b), Fish-Scale type louver
Hole shape fig. 3(d), Bridge type louver
(a)
(b)
(c)
(d)
2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
1000
2000
3000
4000
5000
6000
[1/min]
60
70
80
90
100
[dB(A)]
2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
1000
2000
3000
4000
5000
6000
60
70
80
90
100
2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
1000
2000
3000
4000
5000
6000
60
70
80
90
100
2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 [Hz]
1000
2000
3000
4000
5000
6000
60
70
80
90
100
Hole shape fig. 3(c), circular holes d=3.5 mm
Hole shape fig. 3(c), circular holes d=3.5 mm wi th dimple rings as in fig. 9
Hole shape fig. 3(b), Fish-Scale type louver
Hole shape fig. 3(d), Bridge type louver
(a)
(b)
(c)
(d)

Figure 10: Comparison tail-pipe noise spectra on the roller
dynamometer with V6 engine of different hole shapes (see figure 3) in
Muffler 2 with perforation length L
perf
=L
M
.
In addition different louver shapes have been produced
(see Figure 3). These louvers have also indentations
into the pipe. The Fish-Scale type louver is frequently
used by exhaust manufacturers, whereas Bridge type
is not used in automotive mufflers so far. The idea
behind all these actions is to disturb the formation of a
regular vortex sheet at the perforation.
Figure 10 shows the sound pressure level spectra of all
these perforation types in Muffler 2, in comparison with
the hole perforated muffler measured on the V6 engine.
Note that none of these perforation types whistles, but
they have different broad band flow noise contributions.
As illustrated in the sound pressure level spectrum plot
in Figure 10 (d) the Bridge type perforation has the
lowest flow noise followed by the dimpled perforation
(Figure 10 (b)). The fish-scale type has the highest flow
noise, which can be seen by the high content of green in
the high speed region (Figure 10 (c)). Another view is
displayed in figure 11. Here the overall level is plotted.
The level of the hole perforation is totally dominated by
the whistling and more than 10 dB(A) louder than the
other perforation types. Dimpled perforation and both
louvers are close together with regard to the overall
level. Only at high speed the higher broad band flow
noise of the Fish-Scale type causes 5 dB(A) higher
levels compared to the Bridge type. A major difference
can also be seen in the backpressure, measured in front
of Muffler 2 at 6000 rev/min. The hole perforation has
the lowest backpressure (78 mbar), whereas the Fish-
Scale type has a 260 % higher backpressure (233
mbar). This is due to the large indentation of this louver
with t
FS
= 5 mm into the pipe. This louver pattern avoids
whistling but results in a high flow resistance due to the
increased wall roughness. Such a great perturbation of
the vortex sheet is not necessary as illustrated by the
Bridge type louver and the dimpled hole perforation
which avoids whistling with a much lower backpressure
(87 mbar and 107 mbar respectively) and flow noise
content.
1000 2000 3000 4000 5000 6000
1/min
80
90
100
110
120
dB (A)
Hole
Dimple hole as in fig. 9
Fish-Scale
Bridge
78 mbar
107 mbar
204 mbar
87 mbar
Backpressure
at 6000 rev/min
Perforation
Hole shape
1000 2000 3000 4000 5000 6000
1/min
80
90
100
110
120
dB (A)
Hole
Dimple hole as in fig. 9
Fish-Scale
Bridge
78 mbar
107 mbar
204 mbar
87 mbar
Backpressure
at 6000 rev/min
Perforation
Hole shape
Figure 11: Comparison overall tailpipe noise on the roller dynamometer
with V6 engine of different hole shapes in Muffler 2 with perforation
length L
perf
=L
M
(same configurations as in fig. 10). Solid black line
perforation with circular holes (d=3.5mm), red dotted line perforation as
sketched in fig. 9, green dashed line Fish-Scale (fig. 3(d)) perforation,
blue dash-dotted line Bridge (fig. 3(b)) perforation. Backpressure was
measured in front of the muffler.

CONCLUSION
Several experiments have been performed in order to
understand the whistling of perforated pipes in exhaust
mufflers. The following rules can be drawn from our
investigations.
A certain perforation length is necessary for whistle
which is in the range of the pipe diameter. Higher
frequency whistle tones at higher flow velocity are
excited first. With longer perforation length, the
excitation is strong enough for lower frequency
resonances. The lowest whistle tone will be excited by
the first propagating transverse mode of the muffler. For
an exact prediction of this whistle tone a Finite Element
or Boundary Element calculation of the cavity modes of
the particular muffler is necessary with consideration of
the perforated pipe.
Strouhal numbers of slots are predicted as in literature
whereas circular holes give lower Strouhal numbers.
The Strouhal number range is not depending on
perforation pattern or porosity for a given hole size and
shape. In addition that range is also valid for hot exhaust
gases as in real application on a vehicle. Therefore it is
possible to estimate, depending on the hole size and the
mass flow range of the engine, the critical flow and
frequency regimes of a muffler with perforated pipes.
Without the use of absorption material, it is possible to
suppress whistling by use of shear flow disturbing
elements like little dimples in the pipe or louvered
perforation. All these measures have the drawback of
increased backpressure and higher broad band flow
noise. The best compromise was obtained by the
dimpled pipe and the Bridge type louver.
REFERENCES
1. M.S. Howe, Acoustics of Fluid-Structure
Interaction, Cambridge University Press 1998
2. P.A. Nelson, N.A. Halliwell, P.E. Doak , Fluid
dynamics of a flow excited resonance, Part I:
Experiment, J. Sound and Vib. 78(1) (1981), 15-38
3. P.A. Nelson, N.A. Halliwell, P.E. Doak , Fluid
dynamics of a flow excited resonance, Part II: flow
acoustic interaction, J. Sound and Vib. 91 (1983),
375-402
4. T.D. Mast, A.D. Pierce, Describing function theory
for flow excitation of resonators, J. Acoust. Soc.
Am. 91(1) (1995), 163-172
5. M. Meissner, Aerodynamically excited acoustic
oscillations in cavity resonator exposed to an air jet,
Acta Acustica 88 (2002), 170-180
6. J.C. Bruggemann, A. Hirschberg, M.E.H. van
Dongen, A.P.J. Wijnands, J. Gorter, Self sustained
aero-acoustic pulsations in gas transport system:
analysis of the influence of closed side branches, J.
Sound Vib. 150(3) (1991), 371-393

7. L.F. East, Aerodynamically induced resonance in
rectangular cavities, J. Sound and Vib. 3 (3) (1966),
277-287
8. W.M. Jungowski, K.K. Botros, W. Studzinski,
Cylindrical side-branch as tone generator, J.
Sound and Vib. 131(2) (1989), 265-285
9. A. Selamet, D. Kurniawan, B.D. Knotts, J.M. Novak,
Study of whistles with a generic side branch, SAE
1999-01-1814
10. S. Mallick, R. Shock, V. Yakhot, Numerical
simulation of excitation of a Helmholtz resonator by
grazing flow, J. Acoust. Soc. Am. 114(4) (2003),
1833-1840
11. G.B. Ashcroft, K. Takeda, X. Zhang, A numerical
investigation of the noise radiated by a turbulent flow
over a cavity, J. Sound and Vi. 265(2003), 43-60
12. P.M. Radavich, A. Selamet, J.M. Novak, A
computational approach for flow-acoustic coupling in
closed side branches, J. Acoust. Soc. Am. 109(4)
(2001), 1343-1353
13. J.Y. Chung, D.A. Blaser, Transfer Function Method
of measuring In-Duct acoustic properties. I. Theory
and II. Experiments, J. Acoust. Soc. Am., 68(3)
(1980), 907-921
14. A.J. Green and P.N. Smith, Gas flow noise and
pressure loss in heavy vehicle exhaust systems,
ImechE C17/88 (1988) , 47
15. L. Dedene, M. van Overmeire, P. Guillaume, R.
Valgaeren, Determining the link between the
disturbing sound elements and the construction of
exhaust system, SAE 2001-01-1434
16. L.J. Eriksson, P.T. Thawani, Theory and Practice in
exhaust system design, SAE 850989
17. B.D. Knotts, A. Selamet, Suppression of flow
acoustic coupling in sidebranch ducts by interface
modification, J. Sound and Vib. 265 (2003), 1025-
1045

S-ar putea să vă placă și