Content uploaded by John Kennedy
Author content
All content in this area was uploaded by John Kennedy on Mar 18, 2016
Content may be subject to copyright.
Proceedings of Meetings on Acoustics
Volume 12, 2011 http://acousticalsociety.org/
161st Meeting
Acoustical Society of America
Seattle, Washington
23 - 27 May 2011
Session 5aNS: Noise
5aNS13. The effects of windscreen flow on noise in motorcycle helmets
John Kennedy, Michael Carley*, Nigel Holt and Ian Walker
*Corresponding author’s address: Mechanical Engineering, University of Bath, Claverton Down, Bath, BA2 7AY,
Avon, United Kingdom, m.j.carley@bath.ac.uk
Vortex shedding from a motorcycle windscreen results in three flow regions in which the rider’s helmet may be
immersed. First, the helmet may be completely in the free stream. Second, it may be in the path of vortex shedding from
the windscreen. Third it may be underneath the shed vortices and shielded from the free stream by the windscreen.
On-track noise tests were conducted and showed a difference in sound pressure level of more than 10dB and a change in
spectra content, due to changes in riding position and helmet angle. Similar tests carried out in a wind tunnel, using simul-
taneous microphone and flow visualization measurements, allowed the identification of the flow regions. The potential
contribution of vortex shedding to the noise was assessed using wavelet analysis to identify intermittent flow structures.
Published by the Acoustical Society of America through the American Institute of Physics
Kennedy et al.
© 2011 Acoustical Society of America [DOI: 10.1121/1.3646304]
Received 5 Aug 2011; published 15 Sep 2011
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 1
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
1 Introduction
Noise-induced hearing loss in motorcyclists is a problem which is known to affect professional
riders [1], in particular police officers [2, 3] and racing riders [4]. These previous studies of the
causes and extent of such hearing damage have largely been motivated by the health and safety
implications of hearing impairment and the resulting potential for litigation.
In the late 1980s it was reported [5] that noise levels at the ear of a rider can exceed 90dB(A)
at a speed of 50km/h(14m/s) and can reach 105dB(A) at a speed of 112km/h(31m/s). Given
such high noise exposures, it is no surprise that professional riders have been found to have hearing
loss ranging from 6% in driving instructors to 40% in racing riders [1], where hearing loss is the
percentage of the exposed population that will suffer a reduction in hearing sensitivity of 30dB or
more.
While there have been a number of studies which have measured the noise inside a helmet
few researchers have studied how the interaction between the numerous noise sources, such as the
engine, windscreen and flow around the helmet, combine to generate the at-ear sound. Previous
research [2, 3] has shown that the effect of riding position can change at-ear sound pressure levels
by as much as 9dB. The same authors also highlighted the differences that exist between the
windscreen flow interacting with the cavity region below the chin bar and regions such as the seals
asound the visor. The work reported here is a first step towards a more detailed examination of the
nature and mechanisms of noise generation by windscreen flow interacting with the helmet.
2 Experimental facilities and instrumentation
The helmet used in the laboratory experiments was taken from a series of helmets provided by
manufacturers for noise investigations. As such, the make and model is covered by a confidentiality
agreement. It is a commercially available extra large (XL) motorcycle helmet the dimensions of
which were 26cm ×25cm ×36cm.
The large wind tunnel facility at the University of Bath, Figure 1, was used to test the motor-
cycle helmet in combination with a purpose built windscreen rig. This closed loop facility has a
2m×1.5m×3m test section and provided flow velocities up to 25m/swith a freestream turbulence
intensity of 0.1%. The motorcycle helmet was mounted on a structurally isolated rig capable of
controlling helmet angle relative to the free stream, called the alpha rig. This provides dynamic
control of helmet position while isolating the helmet from any wind tunnel vibrations. The wind
tunnel facility also features a custom built and removable windscreen rig. This rig consists of a
flat 700mm square plate which can be positioned over a range of distances and angles relative to
the helmet, as shown in Figure 2. Smoke can be injected into the flow from the centre point of
the windscreen lip for flow visualisation. The smoke is generated using a Jem ZR12 AL smoke
machine which fills a 0.5m3smoke reservoir from which the smoke is pumped to the windscreen
outlet using a low power air pump.
Microphone measurements were acquired using 1/4inch 130D20 PCB Piezotronics micro-
phones connected to a PCB 442B117 signal conditioner. The microphone data were acquired
using a 16 channel NI DAQ system which comprised of a PC with a NI-PCI-MIO-16E-1 acquisi-
Kennedy et al.
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 2
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
Figure 1: Large wind tunnel facility
tion card and BNC-2090 connector box. The microphones were calibrated using a Larson Davis
CAL200 calibration unit.
A set of on-track tests were conducted at the Llandow Circuit, South Wales. On a closed track
it was possible to control speed and to have the rider filmed from an accompanying car, to record
his head position and angle. The motorcycle used was a 2008 Suzuki GSXF-650 and the helmet
a Shoei Raid II. An additional support vehicle, a Saab 9-5, was used to film the riding conditions
using a camcorder mounted in the near side window. Two GPS units were used to give reference
data on motorcycle position and speed over the course of a test. The first unit was mounted on
the motorcycle dashboard and used by the rider to maintain the test speed along the length of the
straights. The second GPS unit was included in the support vehicle and filmed in the same field
of view as the rider to provide a record of test conditions. Measurements were acquired using a
Edirol R-09 stereo digital recorder and miniature Knowles microphones. The microphones were
calibrated using a reference 1/4inch 130D20 PCB microphone calibrated with the Larson Davis
CAL200 unit.
3 Spectral conditioning techniques
The wind tunnel used has no acoustic treatment and as such there is a risk of signal contamination
by spurious background noise. Also, we wish to compare wind tunnel measurements to data taken
on a motorcycle where there is a contribution to the in-helmet noise from the motorcycle engine
and from environmental sources. In order to extract a ‘helmet-only’ spectrum, one which contains
only the noise due to flow over the helmet, we apply a signal conditioning procedure which has
been used in a number of applications [6, 7] to conditionally remove unwanted contributions to the
output signal.
A model for the system is shown in Figure 4. The output signal p(t)is composed of a sum of
inputs gi(t),i=1,2,.... If we consider a two input problem, where g1(t)is a background noise
contribution to p(t)and g2(t)is the ‘real’ aerodynamically generated noise in the helmet, we wish
to remove from p(t)the part of the signal which is correlated with g1(t). This is readily done using
Kennedy et al.
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 3
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
Figure 2: Windscreen rig. Main dimensions: L= 700mm,30◦<α<90◦,150mm <x<
850mm,−50mm <h<150mm
Figure 3: Llandow Circuit
g1(t)
g2(t)
g3(t)
Σp(t)
Figure 4: System model for partial coherence processing
Kennedy et al.
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 4
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
standard signal processing methods. If we wish to remove the effects of multiple signals, however,
we must take account of possible correlations between them. In this case, if we wish to remove
the contributions of gi(t),i=1,2, leaving the ‘true’ signal due to g3(t), we cannot simply remove
from p(t)the part which is correlated with g1(t)and/or g2(t). Instead, we must decorrelate the
input signals before proceeding.
The method of partial coherence is a systematic technique for performing this decorrelation in
order to rigorously assess the contribution of different sources. If the inputs are uncorrelated, the
coherence function of each with the output signal is:
γ2
ip(f)= |Gip (f)|2
Gii(f)Gpp (f),
where Gpp(f)is the autospectrum of p(t),Gii (f)is the autospectrum of gi(t)and Gip is the corre-
sponding cross-spectrum. The contribution of the ith source to the output can then be removed by
subtracting the correlated part:
Gpp.i =(1−γ2
ip)Gpp .(1)
The notation Gpp.i denotes the spectrum of the signal p(t)with the contribution of the ith input
removed.
If the input signals are correlated, however, this procedure is not valid, as it will lead to a
correlated part being subtracted more than once. In this case, the input signals must be processed
to make them mutually uncorrelated. This is done by using a recursive conditioning procedure,
treating each signal in turn:
Gpp.i =(1−γ2
ip.(i−1)!)Gpp.(i−1)!.(2)
Here, Gpp.(i−1)! is the power spectrum of p(t)with the correlated part of all inputs up to i−1
removed and the partial coherence γ2
ip.(i−1)! given by:
γ2
ip.(i−1)! =|Gip.(i−1)!|2
Gii.(i−1)!Gpp.(i−1)!
(3)
where Gip.(i−1)! is the cross-spectrum with the correlated part of inputs up to i−1removed. The
residual autospectra and cross spectra are given by
Gjk.r!=Gjk.(r−1)! −LrjGjr.(r−1)! (4)
with Lrj the conditioned frequency response function
Lrj =Grj.(r−1)!
Grr.(r−1)!
.(5)
For example, in a three input, single output system, Table 1 shows the conditioning sequence.
At step 1, the auto- and cross-spectra for the inputs and output are generated. At step 2, the spectra
Kennedy et al.
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 5
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
Step Spectra
1G11 G12 G13 G1pG22 G23 G2pG33 G3pGpp
2G22.1G23.1G2p.1G33.1G3p.1Gpp.1
3G23.2! G2p.2! G33.2! G3p.2! Gpp.2!
4G3p.3! Gpp.3!
Table 1: Conditioning sequence for a three-input system
for i≥2and for the output have subtracted from them the part which is correlated with the
first input, generating spectra Gij.1. At step 3, the same procedure is applied for i≥3, using
Equations 2 and 3 and the spectra from step 2. This generates spectra Gij.2!, i.e. spectra which
have had removed from them the part which is correlated with inputs 1 and 2. Finally, at step 4, the
procedure is applied to the third input and to the output, generating the final spectra Gij.3! which
have had removed the effects of all three inputs. The right hand column of Table 1 contains a
set of output autospectra which have had successively removed the contribution from each of the
inputs. To return to the concrete example, if the output signal is an at-ear noise recording and
inputs 1 and 2 are measures of background noise, Gpp.2! is the spectrum of the at-ear noise with the
background noise removed, in other words, an estimate of the ‘true’ aerodynamic noise.
4 Wavelet analysis
A method for the eduction of intermittent events in turbulence based on the wavelet transform has
been developed in work by Farge [8] and Camussi and Guj [9] and applied to turbulent flow in
works such as [10]. This method selects events which result in a non-Gaussian probability density
function (PDF) of the wavelet coefficients. The wavelet cofficients are defined as follows:
w(r, x)= 1
rψ∗(x−x)
ru(x)dx(6)
where ψ∗is the complex conjugate of the wavelet function, u(x)is the signal to be analyzed, and
rcorresponds to a scale.
A qualitative measure of intermittency has been defined by Farge [8]:
I(r, x)= |w(r, x)|2
|w(r, x)|2x
(7)
A second indicator of intermittency is the flatness factor of the wavelet coefficients
F(r)=w(r, x)4x
w(r, x)22
x
=I(r, x)2x(8)
In the technique as applied by Onorato et. al [10] the threshold is set on the Ifunction in order
to produce a flatness factor of 3 at each scale. This in done by an iterative process whereby the
threshold on Iis reduced and the flatness factor recalculated until the target value of 3 is reached.
Kennedy et al.
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 6
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
Table 2: Windscreen configurations
Test case αx h
T0 NA NA NA
T1 40◦300mm -50mm
T2 40◦300mm 0mm
T3 40◦300mm 150mm
T4 90◦300mm -50mm
T5 90◦300mm 0mm
T6 90◦300mm 150mm
T7 49◦810mm 340mm
As a result the method selects as events those regions whose effect on the complete signal is to
make the PDF of the wavelet coefficients strongly non-Gaussian in each scale.
In this work the same technique is applied to the pressure signal acquired using a microphone
positioned at the entrance to the ear canal in order to determine the level of intermittency in the
sound field produced by turbulent flow from the windscreen impinging on the helmet.
5 Track tests
A series of tests were conducted on track for different riding conditions at a constant speed
of 22.2m/s(80km/h). Due to a wind speed of 4.6m/son the day of the track tests the effec-
tive air speed was 17.6m/s(63 kmh). Two microphones were mounted at-ear within the helmet.
The digital recorder continuously acquired data and the rider provided audio cues at the start of
each straight. Three riding positions of fully upright, half forward and fully foward were used and
recorded by the support vehicle. An additional test was conducted where the helmet visor was
partially open.
6 Wind tunnel tests
A series of tests were conducted in the large wind tunnel facilty for comparison with realistic
driving data. Flow speeds of 11.1m/sand 22.2m/s(40 and 80km/h) were chosen for comparison
with on-track data. The effect of helmet angle was investigated over the range of 90◦(fully upright)
to 50◦in 10◦steps. Seven windscreen configurations were tested in addition to a baseline case
without a windscreen in the tunnel. The geometries of the windscreen congifurations are given in
Table 2. The final windscreen configuration, T7, was chosen to recreate the on-track windscreen
geometry as accurately as possible with the current facilities. In configuration T6, the windscreen
vibrated excessively and it was not possible to take reliable data, so results for this case are not
presented.
Kennedy et al.
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 7
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
Table 3: Effect of riding position on SPL
Riding position SPL (dB)
Fully upright 103
Half forward 113
Fully forward 113
Fully upright
visor partially open 106
Table 4: Effect of windscreen configuration on SPL (11.1m/s)
α=40
◦α=90
◦α=49
◦
Helmet angle T0 (dB) T1 (dB)T2(dB)T3(dB) T4 (dB)T5(dB)T6(dB) T7 (dB)
90◦98.55 103.80 106.61 106.56 101.40 100.82 99.53 106.61
80◦103.42 104.45 105.32 105.73 101.55 100.33 98.39 106.79
70◦99.74 106.58 102.21 102.21 103.92 100.84 106.62
60◦99.52 110.38 100.18 103.69 105.95 100.39 108.90
50◦101.49 110.39 99.81 102.21 106.17 98.66 110.41
7 Results
Sound pressure levels at-ear were calculated for the on-track tests and are reported in table 3.
The sound pressure level at-ear was also calculated for all of the windscreen configurations. The
contribution of the wind tunnel was removed using the techniques outlined in section 3 and the
sound pressure level calculated from the conditioned spectra. Tables 4 and 5 show the results for
the two flow speeds tested, 40kmh and 80kmh.
Certain windscreen configurations were placed into two groups to further investigate the effects
of the windscreen. The first group consisted of test cases T0, T1, T2 and T3 for a helmet angle
of 50◦: these were chosen since there was a difference of more than 10dB between at-ear sound
pressure levels. The second group consisted of head angles 90◦,70◦and 50◦for test case T5 since
there was minimal change in sound pressure level despite a significant change in the flow path
around the helmet. The results at a flow speed of 22.2m/s(80km/h) are reported here. Figure 5
shows the flow visualization results for the first group and Figure 6 shows the equivalent results
for the second group. At-ear spectra for the two groups are reported in Figures 7 and 8.
Table 5: Effect of windscreen configuration on SPL (22.2m/s)
α=40
◦α=90
◦α=49
◦
Helmet angle T0 (dB) T1 (dB)T2(dB)T3(dB) T4 (dB)T5(dB)T6(dB) T7 (dB)
90◦110.77 114.27 119.62 119.50 115.68 113.46 120.71
80◦113.73 114.76 118.30 118.76 117.35 114.25 119.96
70◦112.46 115.81 115.94 119.31 119.48 113.77 120.85
60◦111.23 114.34 113.14 116.84 119.17 115.75 120.92
50◦111.74 122.12 113.06 115.60 119.53 114.60 121.22
Kennedy et al.
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 8
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
Figure 5: Flow visualization for helmet angle 50◦, cases T1, T2, and T3
Figure 6: Flow visualization for test case T5, helmet angles 90◦,70◦, and 50◦
10110210
3
60
70
80
90
100
110
120
Fre
q
uenc
y
(
Hz
)
SPL (dB)
T0
T1
T2
T3
Figure 7: At-ear spectra for helmet angle 50◦, test cases T0, T1, T2, and T3
Kennedy et al.
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 9
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
10110210
3
60
70
80
90
100
110
Fre
q
uenc
y
(
Hz
)
SPL (dB)
90
°
70
°
50
°
Figure 8: At-ear spectra for test case T5, helmet angles 90◦,70◦, and 50◦
Wavelet analysis was used in order to investigate the nature of the increase in sound pressure
level and associated spectral peak for test case T1 over T0 at a helmet angle of 50◦. Figure 9 shows
the wavelet coefficients from 0–100Hz for test cases T0 and T1. In order to assess the degree of
intermittency of the sound the intermittency measure outlined in section 4 was calculated. The
results are shown in figure 10.
8 Discussion
The results from the on track test, given in table 3, show that for a relatively small change in head
angle and location there can be an increase of 10dB in the at-ear sound pressure level. This is
comparable to the results obtained during the wind tunnel measurements.
Previous work [2, 3] reports that the interaction of the windscreen flow with the chin cavity
region is the dominant noise source. The data given in Tables 4 and 5 supports this hypothesis
as the highest noise levels for the α=40
◦and the α=90
◦test cases are found for the lowest
windscreen tip locations. However it is important to note that helmet angle was also a strong factor
in determining the sound pressure level for a given windscreen configuration. It is possible that
for the shallower helmet angles a stagnation and recirculation region may form between the chin
cavity and rider neck which may interact very differently with the windscreen flow. The baseline
test case with no windscreen, T0, also showed differences of almost 3dB for a change in helmet
angle at a flow speed of 22.2m/s.
Previous researchers [2, 3] have also noted that while a higher windscreen may reduce noise
by reducing interaction of the windscreen flow with the chin cavity, riders will still choose a lower
windscreen configuration in order not to have to look through the windscreen while riding. The
results from this investigation show that a significant reduction in noise can be achieved for an
identical helmet and windscreen tip location by a change in windscreen angle. The sound pressure
levels shown in Table 5 for T2 and T5 show differences of over 6dB which can be attributed solely
to windscreen angle. These tip locations are of particular interest since they are high enough to
Kennedy et al.
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 10
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
Figure 9: Wavelet coefficients for helmet angle 50◦, test cases T0 and T1
Figure 10: Intermittency helmet angle: 50◦Test Cases: T0 T1
Kennedy et al.
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 11
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
shield the chin cavity region while still allowing the rider to view the road above the windscreen.
The potential change in drag forces produced by a different windscreen angle may be a significant
factor in determining the optimum windscreen angle.
Figures 5 and 7 investigate the effect of the windscreen tip height on the at-ear sound. As
previously stated the lowest tip height, test case T1, produces the highest sound pressure level with
an increase of 6.5–10.5dB over the baseline and other test cases. The spectral results show that this
is due to an increase in low frequency sound below 100Hz with a strong peak in the region of 25
to 30Hz. This is most likely due to an interaction of the windscreen flow with the chin cavity. The
nature of this low frequency sound is further assessed in Figures 9 and 10 which show a comparison
of the raw wavelet coefficients and intermittent events found in the baseline test case and T1. As
shown in the figures this increase in sound is produced by a very regular continuous event and the
only intermittent events detected in this range are related to a sudden short change in amplitude of
the continuous noise source. This is an unexpected result, since there is no obvious intermittency
which can be ascribed to the wake shed from the windscreen impinging on the helmet.
Figures 6 and 8 investigate test case T5 where a minimal change in noise level was produced
regardless of helmet angle. Figure 6 shows that as the helmet angle decreases the windscreen flow
changes from passing over the helmet to impinging on the top surface to directly impacting the
visor and being directed downwards. Despite these changes in the interaction of the windscreen
flow with the helmet the spectral content of the noise remained relatively constant. This is further
evidence of the importance of the interaction with the chin cavity.
Test case T7 attempted to recreate the road geometry as closely as possible within the wind
tunnel. This involved placing the windscreen much further upstream than in the other test cases.
The sound pressure level experienced at-ear for this configuration was significantly higher than
almost all the other cases and was not strongly influenced by helmet angle for the 22.2m/sflow
speed. This may be a result of the fact that the windscreen flow is distributed over a much wider
area in the downstream location of the helmet. This may be an important consideration in any
attempts to reduce noise produced by the interaction of the helmet and windscreen flows.
9 Conclusions
The interaction of the helmet with the windscreen has a significant impact on both the sound
pressure level and spectral content of the at-ear noise.
Differences of over 10dB were seen during both wind tunnel and track tests.
The height of the windscreen tip was a strong factor in determining the resulting sound pressure
level however this was also affected by the windscreen and helmet angle.
Optimising the interaction of the helmet with flow from the windscreen is an area where po-
tentially large noise reductions can be achieved.
Kennedy et al.
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 12
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
References
[1] Chris Jordan, Oliver Hetherington, Alan Woodside, and Harold Harvey. Noise induced hear-
ing loss in occupational motorcyclists. Journal of Environmental Health Research, 3(2):70–
77, 2004.
[2] M. C. Lower, D. W. Hurst, and A. Thomas. Noise levels and noise reduction under motor-
cycle helmets. In Proceedings of Internoise 96, pages 979–982, St Albans, 1996. Institute of
Acoustics.
[3] M. C. Lower, D. W. Hurst, A. R. Claughton, and A. Thomas. Sources and levels of noise
under motorcyclists’ helmets. Proceedings of the Institute of Acoustics, 16(2):319–326, 1994.
[4] A. W. McCombe and J. Binnington. Hearing loss in Grand Prix motorcyclists: occupational
hazard or sporting injury. British Journal of Sports Medicine, 28(1):35–37, 1994.
[5] B. C. Ross. Noise exposure of motorcyclists. Annals of Occupational Hygiene, 33(1):123–
127, 1989.
[6] M. Carley and J. A. Fitzpatrick. Spectral conditioning of propeller noise from broadband
sources. Journal of Sound and Vibration, 238(1):31–49, 2000.
[7] H. W. Esmonde, J. A. Fitzpatrick, H. J. Rice, and F. Axisa. Reduced order modelling of
non-linear squeeze film dynamics. Proceedings of the IMechE, 206:225–238, 1992.
[8] Marie Farge. Wavelet transforms and their applications to turbulence. Annual Review of Fluid
Mechanics, 24:395–457, 1992.
[9] R. Camussi and G. Guj. Orthonormal wavelet decomposition of turbulent flows: intermit-
tency and coherent structures. Journal of Fluid Mechanics, 348:177–199, 1997.
[10] Miguel Onorato, Roberto Camussi, and Gaetano Iuso. Small scale intermittency and bursting
in a turbulent channel flow. Physical Review E, 61(2):1447–1454, 2000.
Kennedy et al.
Proceedings of Meetings on Acoustics, Vol. 12, 040005 (2011) Page 13
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 107.175.211.244 On: Mon, 07 Mar 2016 13:15:15
