Sie sind auf Seite 1von 16

aeroacoustics volume 8 · number 4 · 2009 – pages 301 – 316 301

A note on noise generation by large scale


turbulent structures in subsonic and
supersonic jets
Philip J. Morris*
Department of Aerospace Engineering
The Pennsylvania State University
University Park, PA 16802, U.S.A. pjn@psu.edu

ABSTRACT
This paper describes a method to connect the measured spectral density in the acoustic far field
with the wavenumber/frequency spectrum of the near field fluctuations that produce the noise. A
relationship is first derived between the far field spectral density and the wavenumber/frequency
spectrum of the pressure fluctuations on a cylindrical surface surrounding the jet in the near field.
Measurements of the far field spectral density are then decomposed into contributions from the
Large Scale Similarity (LSS) and Fine Scale Similarity (FSS) spectra. The near field
wavenumber/frequency spectrum associated with the LSS spectral density alone is then
determined. It is shown to have a very similar form for a range of jet operating conditions and
Strouhal numbers. This form is consistent with an instability wave or wave packet model. Since
both subsonic and supersonic jet exit conditions are examined, it is argued that the noise
radiation in the peak radiation directions is controlled by the axial evolution of the turbulent large
scale structures and is not associated with a combination of source convection and mean flow
refraction effects.

1. INTRODUCTION
In spite of sixty years of research into the sources of noise generation in jets, no
consensus yet exists in the aeroacoustics community about the fundamental noise
source mechanisms. In addition, there is still no technique that enables the prediction
of jet noise on the basis of the jet operating conditions and the nozzle geometry alone.
It could be argued that Direct Numerical Simulations (DNS) and Large Eddy
Simulations (LES) have made steps in this direction, but they are computationally
expensive and, by themselves, do not provide an understanding of the noise generation
mechanisms: they are another form of jet noise experiment.

*Boeing/A. D. Welliver Professor


302 A note on noise generation by large scale turbulent structures in
subsonic and supersonic jets

Today, the aeroacoustics research community is faced with two different concepts
for the noise generation and radiation mechanisms. The first methods follow, in some
fashion, the original work of Lighthill [1, 2]. These approaches are called acoustic
analogies and are based on a rearrangement of the equations of motion into the form of
an equation with a wave propagator as the left hand side and the remaining terms acting
as equivalent sources on the right hand side. The effects of source convection were
considered by Ffowcs Williams [3] and the effect of the mean flow on the propagation
of the sound was included in acoustic analogies by Lilley [4] and Goldstein [5]. Many
reviews of the subsequent developments of these approaches are available, including
techniques to combine them with Computational Fluid Dynamics (CFD) to make noise
predictions. In general, these predictions are excellent at large angles to the jet
downstream axis, but not in the peak noise directions at smaller angles to the
downstream jet axis. Examples include predictions by Khavaran and Bridges [6],
Morris and Boluriaan [7], and Tam and Auriault [8]. The last approach is not based on
an acoustic analogy, but is closely related. A notable recent exception is the analysis and
predictions by Goldstein and Leib. [9] The primary differences from previous acoustic
analogy approaches in this study are in the modeling of the source statistical properties
and the use of a propagator that accounts for the slow divergence of the jet flow.
The second point of view argues that the jet noise radiated in the peak noise
radiation directions can be associated with another source mechanism. When the jet
turbulence is convecting supersonically with respect to the ambient speed of sound, a
highly directional noise directivity is observed. Theoretical work by Tam and Morris
[10], Morris and Tam [11] and Tam and Burton [12] and experimental measurements
by McLaughlin et al.[13] have shown that this noise radiation is associated with the
large scale turbulent structures in the jet, and that they can be modeled as linear
instability waves. Tam et al. [14] examined a large database of supersonic jet noise
measurements and showed that the measured far field spectral density could be
collapsed using two similarity spectra. The spectrum shape that dominated in the peak
noise directions was designated as the Large Scale Similarity (LSS) spectrum, being
associated with the large scale structures in the jet shear layer. A second spectrum, the
Fine Scale Similarity (FSS) spectrum, was found to fit the data at larger angles to the
jet downstream axis. At intermediate angles, a combination of the two spectra was
needed to match the experimental measurements. Remarkably, the same two spectra
also permit the collapse of measured spectra at convectively subsonic jet operating
conditions. Additional experimental evidence by Viswanathan [15] and Tam et al. [16]
reinforce the idea that there are two separate noise generation mechanisms in jets at all
operating conditions. It is the latter idea that is pursued in the present paper. The goal
is to determine whether the properties of the LSS spectrum, measured in the far field,
are compatible with a consistent large scale structure model at both subsonic and
supersonic jet operating conditions.
The present paper follows, to some extent, the previous analyses and experiments by
Reba et al. [17], Suzuki and Colonius [18], Ladeinde et al. [19], and Reba et al. [20].
Each of these papers look for a connection between the near and far fields of a jet and
base their analyses on near field models and experiments. In the present paper, no model
aeroacoustics volume 8 · number 4 · 2009 303

for the near field structure is assumed a priori. The behavior is deduced from far field
observations. In addition, only the near field fluctuations that contribute to the LSS
spectrum are extracted. Though, as discussed below, the present approach has
limitations, the consistency of the inferred near field behavior with a traveling wave
packet model for the noise source, at both subsonic and supersonic jet operating
conditions, provides additional evidence for the validity of such a representation.
In the next section, a relationship is developed between the wavenumber spectrum
on a cylindrical surface surrounding the jet and the spectral density in the far field. In
the following section, the far field spectrum is decomposed into the large scale and fine
scale similarity components. These data are then used to determine the magnitude of the
near field wavenumber spectrum associated with the large scale similarity spectrum
alone. Finally, it is shown how these wavenumber spectra conform to a physically
realizable representation of the evolution of large scale structures, modeled as
instability waves or wave packets, in the jet.

2. ANALYTICAL DEVELOPMENT
Consider a cylindrical surface of radius a surrounding an axisymmetric jet. It is
assumed that a is sufficiently close to the jet to sense both the radiating and non-
radiating components of the turbulent pressure fluctuations. It is also assumed that the
cylinder is in a stationary ambient medium surrounding the jet.
In the stationary medium surrounding the jet the pressure fluctuation p(r,ϕ,z,t)
satisfies the three-dimensional wave equation, written here in cylindrical polar
coordinates.

∂2 p 2  ∂2 p 1 ∂p 1 ∂2 p ∂2 p 
− co  2 + + + = 0, (1)
∂t 2  ∂r r ∂r r 2 ∂ϕ 2 ∂z 2 

where (r,ϕ,z) are the coordinates in the radial, azimuthal, and axial directions,
respectively, relative to the jet downstream axis, and co is the speed of sound in the
ambient medium.
The pressure can be represented in terms of its Fourier transform with respect to
time, t and axial distance, and a Fourier series in the azimuthal direction. That is,

∞ ∞ ∞
1
p (r , ϕ, z, t ) = ∑ P (r, k, ω ) exp i (kz + nϕ − ωt ) dωdk,
(2 π ) ∫ ∫
3 n (2)
n =−∞ −∞ −∞

and,
∞ ∞ π

Pn (r , k, ω ) = ∫ ∫ ∫ p(r, ϕ, z, t ) exp[−i(kz + nϕ − ωt )]dϕdzdt,


−∞ −∞ − π
(3)
304 A note on noise generation by large scale turbulent structures in
subsonic and supersonic jets

where, ω is the radian frequency and k is the axial wavenumber. Then, Pn (r, k, ω )
satisfies the equation,

d 2 Pn 1 dPn  ω 2 n2 
+ + − k 2
− P = 0.
dr 2 r dr  co2 r 2  n
(4)

The general solution to Eqn. (4) involving either outgoing or decaying waves can be written,

Pn ( r, k, ω ) = An ( k, ω ) H n(1) ( λr ), (5)

Where H n(1) (ζ) is the Hankel function of the first kind of order n and argument ζ. Also,

λ = (ω 2 / co2 − k 2 )1/ 2 with − π / 2 < arg(λ ) ≤ π /2. (6)

The variation of An (k, ω) is determined by the variation of the pressure on the


cylindrical surface surrounding the jet.
The pressure fluctuations on the cylindrical surface are not simply associated with
a single frequency and are not deterministic. There is a random or stochastic nature to
the pressure field. To account for this, following Tam and Chen [21] and Tam [22], the
wavenumber/frequency spectrum of the n-th azimuthal mode of the pressure
fluctuations on the cylindrical surface is represented by,

Pn ( a, k, ω ) = an (ω )Gn ( k, ω ), (7)

where an(ω) is a real random function of freqeuncy and, as a normalization condition,

∫ Gn (ω, k )dk = 1 (8)


−∞

Then, from Eqns. (5) and (7),

H n(1) ( λr )
Pn ( r, k, ω ) = an (ω ) G ( k, ω ). (9)
H n(1) ( λa ) n

The pressure fluctuation external to and on the cylindrical surface, is given by the
inverse Fourier transforms,
∞ ∞
1 ∞
H n(1) ( λr )
p(r , ϕ, z, t ) =
( 2 π )3
∑ ∫ ∫ a (ω )G (k, ω ) H
n =−∞ −∞ −∞
n n (1 )
( λa )
exp[i( kz + nϕ − ωt )]dωdk. (10)
n
aeroacoustics volume 8 · number 4 · 2009 305

2.1. The far field spectral density


From Eqn. (10), the autocorrelation for the pressure is given by,

∞ ∞
1 ∞ ∞
p (r , ϕ, z, t ) p (r , ϕ, z, t + τ ) =
( 2π )6 n∑ ∑ ∫ ... ∫ an (ω ) an′ (ω ′ ) Gn ( k , ω ) Gn′ ( k ′, ω ′ )
=−∞n ′=−∞−∞ −∞

H n(1) ( λr ) H n(1′ ) ( λ ′r )
× exp
H n(1) ( λa ) H n(1′ ) ( λ ′a ) (11)

exp [ i ( k + k ′ ) z + i ( n + n ′ ) ϕ − i ( ω + ω ′ ) t − iω ′τ ] d ωd ω ′ dkdk ′,

where the overbar denotes an ensemble average and λ ′ = (ω' / c o − k ′ ) .


2 2 2 1/ 2

To provide a definition of the properties of the random variable, let the


autocorrelation of the pressure on the cylindrical surface at z = 0 be given by, (see Tam
[21] ),

p( a, ϕ, 0, t ) p( a, ϕ + χ, 0, t, + τ ) = A ρj U j D j δ( τ )δ( χ).
2 2 3
(12)

This is equivalent to stating that there are no preferred frequencies or azimuthal mode
numbers: the initial pressure fluctuations have a “white noise” autospectrum at z = 0.
Here, A is a non-dimensional constant, and Dj, ρj, and Uj are the jet exit diameter,
density, and velocity, respectively. From Eqns. (11) and (12), as well as the
normalization condition, Eqn.( 8), it can be shown that,
(13)
an (ω )an′ (ω ′ ) = A ( 2 π) ρj U j D j δ(ω + ω ′ )δ n,− n′ .
2 4 2 3

From Tam [22] (Appendix A), noting the slightly different definitions of variables and
the associated choice of branch cuts, it can be shown that,

λ ( − k , − ω ) = − λ * (k , ω ) (14)

where an asterisk denotes a complex conjugate. Also, from the analytic continuation
properties of Hankel functions,

*
H −(1n) ( − λ* r ) = −  H n(1) ( λr )  .
(15)

From its definition, it is also readily seen that,

G− n ( − k, −ω ) = Gn* ( k, ω ). (16)

It is also convenient to introduce the function Fn(r, z,ω ) given by,


306 A note on noise generation by large scale turbulent structures in
subsonic and supersonic jets

∫ G (k, ω) HH [λ ( k , ω ) r ]
(1 )
(17)
Fn (r , z, ω ) = n
exp (ikz) dk.
–∞
n (1 )
n
[λ ( k , ω ) a]

Substitution of these relationships into Eqn. (11), following integration with respect to
ω′ , and a change of the variable k′ to –k′, yields,


A2 2 3
p(r , ϕ, z, t ) p(r ,ϕ , z, t + τ ) = ρ U D ∑
( 2 π)2 j j j n=−∞ ∫–∞
Fn ( r , z , ω ) 2 exp(– i ωτ) dω. (18)

The spectral density of the pressure fluctuations is given by,


S (r, ϕ , z, ω) = ∫ (p (r , ϕ, z,t) p(r , ϕ, z, t + τ) exp (iωτ ) d τ.


–∞
(19)

Then the integration with respect to τ can be performed and the spectral density is found
to be,

A2 2 3
ρj U j D j ∑ Fn (r, z, ω ) .
2
S (r , z,ω ) = (20)
2π n −∞

Before proceeding, a simplifying assumption will be made to only consider the n = 0


case. This should be an adequate first step for the axisymmetric jet case. On a practical
level, the far field measurements used in the present paper are time-averaged and only
made at one azimuthal angle. So, any attempt to extract the separate contributions from
the different azimuthal modes would require a much more comprehensive database. A
similar assumption was made by Tam [22] in his stochastic model of broadband shock-
associated noise. So, the subsequent analysis can be considered to give an azimuthally
averaged result.
Equation (20) is valid for any value of r and z, (and ϕ), external to the cylindrical
surface. However, the measurements used in the present analysis are performed in the
far field. To obtain an expression for the spectral density in the far field, spherical polar
coordinates are introduced, such that

z = R cos θ, r = R sin θ. (21)

Also, in the limit of large argument, the Hankel function takes the asymptotic form,

2 (22)
H o(1) (ζ) → exp[i (ζ − π / 4 )] as ζ → ∞ .
πζ

Then, the method of stationary phase can be used to evaluate F (R, θ, ω). This gives,
aeroacoustics volume 8 · number 4 · 2009 307

2i G (ω cos θ / co , ω ) (23)
F ( R, θ, ω ) = − exp(iω R / co ).
R H o(1) (ω a sin θ / co )

Finally, using Eqns. (22) and (23), a relationship can be derived between the non-
dimensional wavenumber/frequency spectrum of the pressure fluctuations on the
cylindrical surface and the far field spectral density. It is given by,

2 2
G (ω cos θ / co , ω ) π R  S ( R, θ, ω ) (24)
A 2
=   H o(1) (ωa sin θ / co ) 2
,
D 2j 2  Dj  ρj2U 3j D j

Note that Eqn. (24) only provides the near field wavenumber spectrum for
wavenumbers given by, |k| ≤ ω/co.
In the next section, the measured far field spectral density is analyzed to extract the
component associated with the large scale structures. The resulting component is then
used in Eqn. (24) to determine the wavenumber/frequency spectrum of the near field
pressure fluctuations on the cylindrical surface that give rise to this observed far field
behavior. As noted already, it is clear from Eqn. (24) that only the supersonic phase
velocity or radiating components of the near field pressure fluctuations can be extracted
from the far field data. However, this does not mean that the method is only applicable to
jets with supersonically convecting turbulent structures, since the turbulent fluctuations at
any jet exit velocity will contain some components with supersonic phase speeds.

3. ANALYSIS OF EXPERIMENTAL MEASUREMENTS


The experiments were conducted in the Aeroacoustic Propulsion Facility at the NASA
Glenn Research Center. The measurements were made with both converging and
converging-diverging nozzles. All the nozzles had a 2 inch exit diameter. The data have
been corrected to be lossless and the observer radius is at 100 jet diameters.
Descriptions of the test facility and the converging nozzle are given by Brown and
Bridges [23]. The test cases chosen are summarized in Table 1.
Nozzle C000 is a converging nozzle, and nozzles C015 and C018 are converging-
diverging (CD) nozzles with design Mach numbers of 1.4 and 1.8 respectively. The total
temperature ratios have been selected so that the jet exit static temperature ratio is unity,
with the exception of the lowest velocity case where the jet is unheated. The CD nozzles
are operated on design.
The first step in the data analysis process is to separate the total spectrum into its
large scale similarity (LSS) and fine scale similarity (FSS) spectra, as developed by
Tam et al. [14]. Tam et al. [14] interrogated a large data base of acoustic measurements
made in the Jet Noise Laboratory at the NASA Langley Research Center. As noted in
the Introduction, they showed that the experimental data could be collapsed in terms of
two universal similarity spectra. They argued that the LSS spectrum, that fits the
measured spectra in the peak noise direction, is associated with the large scale turbulent
structures in the jet. Since they were examining supersonic data, and it is known that
the large scale turbulent structures generate instability wave noise radiation in the peak
308 A note on noise generation by large scale turbulent structures in
subsonic and supersonic jets

Table 1: Nozzles and operating conditions.


Design Nozzle Total Jet exit
Nozzle ID mach number pressure ratio temperature ratio velocity, Uj m/s
C000 1.0 1.1916 1.000 171
C000 1.0 1.6883 1.160 308
C015 1.4 3.1820 1.392 480
C018 1.8 5.7450 1.648 618

10

0
Spectral density, dB relative to peak

−10

−20

LSS spectrum
FSS spectrum
−30

−40 −2
10 10−1 100 101
Frequency, f/fpeak

Figure 1: LSS and FSS similarity spectra.

noise direction, this is a reasonable assumption. Subsequently, several authors have


shown that the same LSS spectrum fits data at subsonic speeds as well (see
Viswanathan [15,24] , for example). The FSS spectrum fits the acoustic data at larger
angles to the jet downstream axis. The two similarity spectra are shown in Fig. 1.
Best fits, based on the two similarity spectra, have been determined for the spectral
density data for each operating condition. Many of the features of the data described by
Tam et al.[16] were noticed during this analysis. For example, the peak Strouhal
number for the FSS component varies only slightly with observer polar angle, whereas
the peak Strouhal number for the LSS component drops rapidly as the observer angle
moves closer to the jet downstream axis. However, here, only a few examples of the
aeroacoustics volume 8 · number 4 · 2009 309

Spectral density, dB re 2×10-5 Pa/unit St

Spectral density, dB re 2×10-5 Pa/unit St


90 90

80 80

70 70

60 60
Experiment Experiment
Total spectrum Total spectrum
50 LSS spectrum 50 LSS spectrum
FSS spectrum FSS spectrum

40 40
10−2 10−1 100 101 10−2 10−1 100 101
Strouhal number, St = fDj/Uj Strouhal number, St = fDj/Uj
(a) θ = 45°, Uj = 171 m/s. (b) θ = 25°, Uj = 171 m/s.
Spectral density, dB re 2×10-5 Pa/unit St

Spectral density, dB re 2×10-5 Pa/unit St


140 140

130 130

120 120

110 110
Experiment Experiment
Total spectrum Total spectrum
100 LSS spectrum 100 LSS spectrum
FSS spectrum FSS spectrum

90 90
10−2 10−1 100 101 10−2 10−1 100 101
Strouhal number, St = fDj/Uj Strouhal number, St = fDj/Uj
(c) θ = 45°, Uj = 618 m/s. (d) θ = 25°, Uj = 618 m/s.

Figure 2: Decomposition of the spectral density into LSS and FSS similarity
components.

decomposition are given. Fig. 2 shows the decomposition of the spectra at 25 and 45
degrees to the jet downstream axis for the lowest and highest speed cases as a function
of Strouhal number, St = f Dj /Uj . Note that in Fig. 2 (d) no significant contribution has
been assumed to be made by the FSS spectrum. Thus, the LSS and total spectra are
equal.
It should be noted that there is more than a little room for discretion in generating
the spectral decomposition. The peak frequency and level for both similarity spectra
need to be determined. In most cases, a perfect fit to the experimental data at all
frequencies is not possible. Some of this is due to the original similarity spectra being
generated for experimental data that was not corrected for atmospheric absorption:
310 A note on noise generation by large scale turbulent structures in
subsonic and supersonic jets

2.5

LSS spectral density, Paˆ2/Hz


LSS spectral density, Paˆ2/Hz
2E-05 2

1.5

1E-05 1

0.5

0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Angle to downstream axis θ, degrees Angle to downstream axis θ, degrees
(a) Uj = 171 m/s. (b) Uj = 618 m/s.

Figure 3: LSS spectral densities. St = 0.1.

whereas the present data has been made lossless. In addition, Viswanathan[24] has
shown that the jet total temperature ratio affects the LSS spectral shape: particularly at
higher frequencies. However, the focus of the present study is at Strouhal numbers close
to the peak noise Strouhal number: in fact Strouhal numbers of 0.05, 0.1, 0.2 and 0.4
have been selected. In addition, only information concerning the LSS similarity
spectrum has been used and its peak frequency and level are easier to identify,
especially at relatively small angles to the jet downstream axis, where it dominates the
radiated noise.
Once the measured spectra at all angles have been decomposed, the spectral density
at the selected Strouhal numbers as a function of observer angle can be determined.
Then the near field wavenumber spectra at those Strouhal numbers can be calculated.
This process is described in the next section.

4. NEAR FIELD LSS WAVENUMBER SPECTRUM


In the previous section a method by which the variation of the LSS similarity spectral
density with observer angle can be determined was described. In this section this
information is used in Eqn. (24) to determine the wavenumber spectra at selected
frequencies on the cylindrical surface in the near field.
Figure 3 shows the LSS spectral density for Strouhal number St = 0.1 as a function
of observer angle for the lowest and highest speed cases. The symbols show the
observer angles at which the data were taken with a spacing of 5 degrees. The data at θ
= 15 degrees are sometimes affected by the jet flow field itself. But this data has been
included in all the cases. It should be noted that the spectral density is shown with a
linear scale, so it represents S (R, θ, ω) in Pa2/Hz. Clearly the levels associated with the
highest velocity case are much greater.
These spectra, for all four jet exit velocity cases and four Strouhal numbers, are used
in the evaluation of the right hand side of Eqn. (24). The remaining terms are all known
quantities, so that the wavenumber spectra can be evaluated. The quantity AG (ωcos
θ/co,ω) is shown for all the cases in Fig. 4. The data conform to a remarkably consistent
aeroacoustics volume 8 · number 4 · 2009 311

Wavenumber spectrum amplitude, A/G(ωcosθ/co,θ)/Dj

Wavenumber spectrum amplitude, A/G(ωcosθ/co,θ)/Dj


0.001 Uj = 618 m/s 0.001 Uj = 618 m/s
Uj = 480 m/s
Uj = 480 m/s
Uj = 308 m/s
Uj = 308 m/s
Uj = 171 m/s
Uj = 171 m/s

0.0005 0.0005

0 0
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0.8 1 1.2
Wavenumber, kDj Wavenumber, kDj
(a) St = 0.05. (b) St = 0.10.
Wavenumber spectrum amplitude, A/G(ωcosθ/co,θ)/Dj

Wavenumber spectrum amplitude, A/G(ωcosθ/co,θ)/Dj

Uj = 618 m/s
0.001 0.001 Uj = 618 m/s
Uj = 480 m/s
Uj = 480 m/s
Uj = 308 m/s
Uj = 308 m/s
Uj = 171 m/s
Uj = 171 m/s

0.0005 0.0005

0 0
0 0.5 1 1.5 2 2.5 0 1 2 3 4 5
Wavenumber, kDj Wavenumber, kDj
(c) St = 0.20. (d) St = 0.40.

Figure 4: Wavenumber spectra of pressure fluctuations on a cylindrical surface of


radius D j.

pattern. As the jet exit velocity increases so the levels of the wavenumber spectra at a
given wavenumber decrease slightly. But all the spectra follow the same pattern with
increasing wavenumber that is shown in the most complete manner at the highest
velocity. It should be noted that the most reliable values of the LSS spectral density
correspond to higher end of the wavenumber range for each velocity condition, as this
312 A note on noise generation by large scale turbulent structures in
subsonic and supersonic jets

Table 2: Estimated convection velocity as a function of Strouhal number.

St kpeakDj Uc/Uj
0.05 0.533 0.59
0.10 1.07 0.59
0.20 1.93 0.65
0.40 3.71 0.68

is related to the smaller observer angles to the jet downstream axis where the LSS
spectrum dominates. Values at the lower end of the wavenumber range should be
regarded with more suspicion as they are related to the approximate decompositions,
examples of which are shown in Fig. 2, where both FSS and LSS spectra contribute
almost equally to the total spectrum. It should be remembered that only those
wavenumbers that have a sonic phase velocity in the direction of a far field observer,
can actually radiate noise. This is discussed further in the next section, where the
wavenumber spectra are analyzed and the implications for the noise source mechanism
are considered.

5. DISCUSSION
In the case of a jet where the turbulence is convecting supersonically with respect to the
ambient speed of sound, it has been shown that the large scale structures, modeled as
instability waves, radiate directly to the far field (see Morris and Tam [11] and Tam and
Burton [12]). As the instability wave travels downstream, its amplitude increases and
eventually decays. A very simple model consists of a wave packet with a Gaussian
amplitude distribution and a constant phase velocity. The wavenumber spectrum
associated with this simple model has a peak at a wavenumber given by kpeak = ω/Uc
where ω is the radian frequency of the instability wave and Uc is its phase velocity. The
width of the wavenumber spectrum has an inverse dependence on the width of the
Gaussian envelope. This simple model can be used to examine the wavenumber spectra
determined in the previous section.
Consider the highest velocity case. The wavenumber peak can be estimated from the
spectra and the radian frequency is determined from the Strouhal number. The inferred
phase velocity is shown in Table 2. These are quite realistic numbers based on
experimental measurements of turbulence convection velocities. It should be
remembered that there are some potential inaccuracies in determining the LSS spectrum
and also in estimating the peak wavenumber. However, the trend of increasing phase
velocity with increasing Strouhal number is very encouraging. It should be noted that
in the two lowest Strouhal number cases the convective Mach number is only just
supersonic. This is consistent with the wavenumber spectra ending very close to the
peak.
Before considering some additional implications of the wavenumber spectra shown
in Fig. 4, some general comments are relevant. As mentioned in the last section, the
spectra from a wide range of operating conditions conform to a systematic pattern on a
aeroacoustics volume 8 · number 4 · 2009 313

linear scale. The same level of pressure fluctuations, relative to mean dynamic pressure
of the jet ρjUj 2, result in radiated noise levels, as shown in Fig. 3 that differ by five
orders of magnitude. This difference in levels can be attributed to the wavenumber
content of the near field pressure fluctuations, since the jet’s mean dynamic head only
varies by a factor of approximately 13 over the range of conditions considered.
Now consider the hypothesis that the wavenumber spectra associated with the lower
velocity cases would have the same amplitude variation with wavenumber as the
highest velocity case. Of course, on the basis of the far field measurements alone, the
peak in the spectra would not be observed, since this wavenumber would be associated
with subsonic phase velocities. However, the peak in the near field wavenumber
spectrum would be expected to remain relatively independent of the jet exit velocity,
since,

Uc ω 2 πSt (25)
= = .
U j k peakUj D j k peak

Though this ratio is known to be relatively insensitive to jet exit velocity a gradual
decrease in the ratio with increase in jet exit velocity would provide an improved
collapse of the near field wavenumber spectra shown in Fig. 4.
There are additional reasons why a perfect collapse of the wavenumber spectra would
not be expected, if ideas from linear instability analysis are invoked. For example, as the
jet exit Mach number increases, so the rate of growth of the jet shear layer decreases. This
would provide a greater axial distance for the instability waves or large scale structures to
develop. At the same time, the rate of growth of the instabilities waves is known to decrease
as the jet Mach number increases, which would offset the effect of the decreasing shear
layer growth rate. Finally, the choice of a cylindrical control surface, though convenient, is
a less than ideal choice, since the distance from the edge of the jet to the control surface,
in terms of jet shear layer thicknesses, is a strong function of axial distance. An improved
formulation would involve a conical control surface as used by Reba et al. [20] or the use
of the method of matched asymptotic expansions as employed by Tam and Burton [12].
Even with these provisos, it can be argued that the observations are consistent with
the same fundamental noise generation mechanism by the large scale structures in the jet
being present at subsonic as well as supersonic convective Mach numbers. The essential
difference in the radiated noise levels is associated with their wavenumber content. At
high convective Mach numbers, the most energetic wavenumber components radiate
directly to the far field. The peak noise direction is controlled by their phase velocity:
being given by θpeak = cos–1(Co /Uc). At convectively subsonic conditions, only the low
wavenumber components in the tail of the wavenumber spectrum radiate noise.
However, the peak noise direction would not be directly downstream, θ = 0, since the
weighting effect of the Hankel function, 1/Ho(1) ( ω a sin θ/Co ) causes the far field levels
to reduce as the downstream axis is approached. Some simple estimates indicate that this
effect causes the peak in the noise spectrum to have a maximum at approximately 30
degrees to the jet downstream axis for all convectively subsonic conditions.
314 A note on noise generation by large scale turbulent structures in
subsonic and supersonic jets

Clearly, additional theoretical work and analysis of experimental data is necessary.


A key difficulty with using far field experimental data is the limit that it places on the
range of near field wavenumbers that can be deduced, especially at the lower velocities.
But, it is in the far field alone that, at the moment, it is straightforward to separate the
contribution of the LSS spectrum to the total radiated noise. However, given these
difficulties and questions, in the author’s opinion, this paper has provided additional
evidence, supported by experimental data, that the noise radiation in the peak noise
directions is generated by the large scale turbulent structures in the jet. And this applies
to both subsonic and supersonic jet conditions.

ACKNOWLEDGMENTS
I am very grateful to Dr. James Bridges and his experimental support team at the NASA
Glenn Research Center for supplying the acoustic data in electronic form - already
corrected for facility and atmospheric factors. I would also like to acknowledge the
many discussions I have had with Dr. K. Viswanathan of the Boeing Company on the
topic of this paper. Some of these ideas have been bounced around with Dr. Dennis
McLaughlin, my colleague at Penn State, who has always emphasized the importance
of experimental data in theoretical modeling. Finally, I would like to acknowledge the
many contributions by Dr. Christopher Tam. I have made use of some of his analysis
and hopefully some of his insight.

REFERENCES
[1] Lighthill, M. J., “On sound generated aerodynamically: I. General theory,”
Proceedings of the Royal Society of London, Vol. A 211, 1952, pp. 564–587.
[2] Lighthill, M. J., “On sound generated aerodynamically. II. Turbulence as a source
of sound,” Proceedings of the Royal Society of London, Vol. A 222, 1954, pp.
1–32.
[3] Ffowcs Williams, J. E., “The noise from turbulence convected at high speed,”
Philosophical Transactions of the Royal Society of London, Vol. A 255, 1963, pp.
469–503.
[4] Lilley, G. M., “On the noise from jets,” Noise Mechanisms, AGARD-CP-131,
1974, pp. 13.1–13.12.
[5] Goldstein, M. E., Aeroacoustics, McGraw-Hill, New York, 1976.
[6] Khavaran, A. and Bridges, J. E., “Modelling of fine-scale turbulence mixing
noise,” Journal of Sound and Vibration, Vol. 279, 2005, pp. 1131–1154.
[7] Morris, P. J. and Boluriaan, S., “The Prediction of Jet Noise from CFD Data,”
AIAA Paper 2004–2977, 2004.
[8] Tam, C. K. W. and Aurialt, L., “Jet mixing noise from fine-scale turbulence,”
AIAA Journal, Vol. 37, No. 2, 1999, pp. 145–153.
aeroacoustics volume 8 · number 4 · 2009 315

[9] Goldstein, M. E. and Leib, S. J., “The aeroacoustics of slowly diverging


supersonic jets,” Journal of Fluid Mechanics, Vol. 600, 2008, pp. 291–337.
[10] Tam, C. K. W. and Morris, P. J., “The radiation of sound by the instability waves
of a compressible plane turbulent shear layer,” Journal of Fluid Mechanics, Vol.
98, No. 2, 1980, pp. 349–381.
[11] Morris, P. J. and Tam, C. K. W., “Near and far field noise from large-scale
instabilities of axisymmetric jets,” AIAA Paper 77–1351, 1977.
[12] Tam, C. K. W. and Burton, D. E., “Sound generation by the instability waves of
supersonic flows. Part 2. Axisymmetric jets.” Journal of Fluid Mechanics, Vol.
138, 1984, pp. 273–295.
[13] McLaughlin, D. K., Morrison, G. L., and Troutt, R. R., “Experiments on the
instability waves in a supersonic jet and their acoustic radiation,” Journal of Fluid
Mechanics, Vol. 69, 1975, pp. 73–95.
[14] Tam, C. K. W., Golebiowski, M., and Seiner, J. M., “On the two components
of turbulent mixing noise from supersonic jets,” AIAA Paper 96–1716, 1996.
[15] Viswanathan, K., “Analysis of the two similarity components of turbulent mixing
noise,” AIAA Journal, Vol. 40, 2002, pp. 1735–1744.
[16] Tam, C. K. W., Viswanathan, K., Ahuja, K. K., and Panda, J., “The sources of jet
noise: experimental evidence,” AIAA Paper 2007–3641, 2007.
[17] Reba, R., Narayanan, S., Colonius, T., and Suzuki, T., “Modeling jet noise from
organized structures using near-field hydrodynamic pressure,” AIAA Paper
2005–3093, 2005.
[18] Suzuki, T. and Colonius, T., “Instability waves in a subsonic round jet detected
using a near-field phased microphone array,” Journal of Fluid Mechanics, Vol.
565, 2006, pp. 197–226.
[19] Ladeinde, F., Cai, X., Alabi, K., Reba, R., Schlinker, R. H., and Simonich, J., “On
the connection between near-field and far-field solutions of high-speed jet noise,”
AIAA Paper 2008–0011, 2008.
[20] Reba, R., Simonich, J., and Schlinker, R., “Measurements of source wave-packets
in high speed jets and connection to far-field sound,” AIAA Paper 2008–2891, 2008.
[21] Tam, C. K. W. and Chen, K. C., “A statistical model of turbulence in two-
dimensional shear layers,” Journal of Fluid Mechanics, Vol. 92, 1979, pp. 303–326.
[22] Tam, C. K. W., “Stochastic model theory of broadband shock associated noise from
supersonic jets,” Journal of Sound and Vibration, Vol. 116, No. 2, 1987, pp. 265–302.
[23] Brown, C. and Bridges, J., “Small hot jet acoustic rig validation,” TM
2006–214234, NASA, 2006.
[24] Viswanathan, K., “Aeroacoustics of hot jets,” Journal of Fluid Mechanics, Vol.
516, 2004, pp. 39–82.

Das könnte Ihnen auch gefallen