You are on page 1of 10

CHAPTER 10

NONLINEAR ACOUSTICS
Oleg V. Rudenko

Blekinge Institute of Technology


Karlskrona, Sweden
Malcolm J. Crocker
Department of Mechanical Engineering
Auburn University
Auburn, Alabama
1 INTRODUCTION
In many practical cases, the propagation, reection,
transmission, refraction, diffraction, and attenuation of
sound can be described using the linear wave equation.
If the sound wave amplitude becomes large enough,
or if sound waves are transmitted over considerable
distances, then nonlinear effects occur. These effects
cannot be explained with linear acoustics theory. Such
nonlinear phenomena include waveform distortion
and subsequent shock front formation, frequency
spreading, nonlinear wave interaction (in contrast to
simple wave superposition) when two or more waves
intermingle, nonlinear attenuation, radiation pressure,
and streaming. Additionally in liquids, cavitation and
water hammer and even sonoluminescence can occur.
2 DISCUSSION
In most noise control problems, only a few nonlinear
effects are normally of interest and these occur either
rst in intense noise situations, for example, close to
jet or rocket engines or in the exhaust systems of
internal combustion engines, or second in sound prop-
agation over great distances in environmental noise
problems. The rst effect the propagation of large
amplitude sound wavescan be quite pronounced,
even for propagation over short distances. The sec-
ond, however, is often just as important as the rst,
and is really the effect that most characterizes nonlin-
ear sound propagation. In this case nonlinear effects
occur when small, but nite amplitude waves travel
over sufciently large distances. Small local nonlinear
phenomena occur, which, by accumulating effects over
sufcient distances, seriously distort the sound wave-
form and thus its frequency content. We shall mainly
deal with these two situations in this brief chapter. For
more detailed discussions the reader is referred to sev-
eral useful and specialized books or book chapters on
nonlinear acoustics.
19
The second effect described, waveform distortion
occurring over large distances, has been known for a
long time. Stokes described this effect in a paper
10
in

Present address: Department of Acoustics, Physics Faculty,


Moscow State University, 119992 Moscow, Russia.
o a b
c
z
Figure 1 Wave steepening predicted by Stokes.
10
1848 and gave the rst clear description of waveform
distortion and steepening. See Fig. 1.
More recent theoretical and experimental results
show that nonlinear effects cause any periodic dis-
turbance propagating though a nondispersive medium
to be transformed into a sawtooth one at large dis-
tances from its origin. In its travel through a medium,
which is quadratically nonlinear, the plane wave takes
the form of a saw blade with triangular teeth.
The transformation of periodic wave signals into saw-
tooth signals is shown in Fig. 2a. As the distance, x,
t
2
1
t t
(a)
2
1
(b)
x = 0 x
1
> 0 x
2
> x
1
(c)
Figure 2 Examples of wave steepening fromRudenko.
22
159
160 FUNDAMENTALS OF ACOUSTICS AND NOISE
from the origin of the sound signal increases, any ne
details in the initial wave prole become smoothed
out through dissipation, during the wave propagation.
The nal wave prole is the same for both a sim-
ple harmonic signal (curve 1) and a more complicated
complex harmonic signal (curve 2) at some distance
from the source (x = x
2
in Fig. 2a).
A single impulsive sound signal becomes trans-
formed into an N-wave (Fig. 2b) at large distances
from its origin if the medium is quadratically non-
linear. Note that the integral of the time history of
the function tends to zero as x as a result of
diffraction.
In cubically nonlinear media the teeth of the saw
blade have a trapezoidal form (Fig 2c). Each wave
period has two shocks, one compression and the other
rarefaction. The existence of sawtooth-shaped waves
other than those shown in Fig. 2 is possible in media
with more intricate nonlinear dissipative and dispersive
behaviors. The disturbances shown in Fig. 2, however,
are the most typical.
These effects shown in Fig. 2 can be explained
using very simple physical arguments.
11
Theoretically,
the wave motion in a uid in which there is an
innitesimally small disturbance, which results in a
sound pressure uctuation eld, p, can be described
by the well-known wave equation:

2
p
x
2

1
c
2
0

2
p
t
2
= 0 (1)
Theoretically, sound waves of innitesimally small
amplitude travel without distortion since all regions of
the waveform have the same wave speed dx/dt = c
0
.
However, even in a lossless medium (one theoretically
without the presence of dispersion), progressive waves
of small but nite amplitude become distorted with
distance and time. This is because, in the regions
of the wave with positive sound pressure (and thus
positive particle velocity), the sound wave travels
faster than in the regions with negative sound pressure
(and thus particle velocity). This effect is caused by
two phenomena
11
:
1. The sound wave is traveling through a velocity
eld consisting of the particle velocity u. So with
waves of nite amplitude, the wave speed (with respect
to a xed frame of reference) is
dx/dt = c +u (2)
where c is the speed of sound with respect to the uid
moving with velocity u.
2. The sound speed c is slightly different from the
equilibrium sound speed c
0
. This is because where the
particle velocity is positive (so is the sound pressure)
the gas is compressed and the absolute temperature T
is increased. Where the particle velocity is negative
(and the sound pressure is too) then the temperature
is decreased. An increased temperature results in a
slightly higher sound speed c and a decreased tem-
perature results in a slightly decreased sound speed c.
Mathematically we can show that the speed of
sound is given by
c = c
0
+[( 1)/2]u (3)
where is the ratio of specic heats of the gas. We
can also show that the deviation of c from c
0
can
be related to the nonlinearity of the pressuredensity
relationship. If Eqs. (2) and (3) are combined, we
obtain
dx/dt = c
0
+u (4)
where is called the coefcient of nonlinearity and is
given by
= ( +1)/2 (5)
The fact that the sound wave propagation speed
depends on the local particle velocity as given by
Eq. (4) shows that strong disturbances will travel faster
than those of small magnitude and provides a simple
demonstration of the essential nonlinearity of sound
propagation.
We note in Fig. 3 that u
p
is the particle velocity
at the wave peak, and u
v
is the particle velocity at
the wave valley. The time used in the bottom gure
of Fig. 3 is the retarded time = t x/c
0
, which
is used to present all of the waveforms together for
comparison. The distance x in Fig. 3c is the distance
needed for the formation of a vertical wavefront.
Mathematically, nonlinear phenomena can be
related to the presence of nonlinear terms in analytical
models, for example, in wave equations. Physically,
nonlinearity leads to a violation of the superposition
principle, and waves start to interact with each other.
As a result of the interaction between frequency
components of the wave, new spectral regions appear
and the wave energy is redistributed throughout
the frequency spectrum. Nonlinear effects depend on
the strength of the wave; they are well-dened if the
intensity of the noise, the amplitude of the harmonic
signal, or the peak pressure of a single pulse is large
enough.
The interactions of intense noise waves can
be studied by the use of statistical nonlinear
acoustics.
1,4,14,16
Such studies are important because
different sources of high-intensity noise exist both
in nature and engineering. Explosive waves in
the atmosphere and in the ocean, acoustic shock
pulses (sonic booms), noise generated by jets,
and intense uctuating sonar signals are examples
of low-frequency disturbances for which nonlinear
phenomena become signicant at large distances.
There also exist smaller noise sources whose
spectra lie in the ultrasonic frequency range. These
include, for instance, ordinary electromechanical
transducers whose eld always contains uctuations,
and microscopic sources like bubble (cavitation) noise
and acoustic emission created during growth of cracks.
Finally, intense noise of natural origin exists, such as
thunder and seismic waves. There are obvious links
between statistical nonlinear acoustics and nonwave
NONLINEAR ACOUSTICS 161
U
U
C
0
+ U
p
C
0
C
0
Waveform in Space
Waveform in Time
C
0
C
0
+ U
p
1
2
t
t
t = t x/c
0
t t
(a) x = 0 (b) x > 0
C
0
- U
v
(d) x > x
x
(c) x = x
Figure 3 Wave steepening predicted by Eq. (4).
20
problemsturbulence, aeroacoustic interactions, and
hydrodynamic instabilities.
3 BASIC MATHEMATICAL MODELS
3.1 Plane Waves
We shall start by considering the simple case of a plane
progressive wave without the presence of reections.
For waves traveling only in the positive x direction,
we have from Eq. (1), remembering that p/u =
0
c
0
,

2
u
x
2

1
c
2
0

2
u
t
2
= 0 (6)
Equation (6) may be integrated once to yield a rst-
order wave equation:
u
x
+
1
c
0
u
t
= 0 (7)
We note that the solution of the rst-order Eq. (7)
is u = f (t x/c
0
), where f is any function. Equa-
tion (7) can also be simplied further by transforming
it from the coordinates x and t to the coordinates x and
, where = t x/c
0
is the so-called retarded time.
This most simple form of equation for a linear traveling
wave is u(x, )/x = 0. This form is equivalent to
the form of Eq. (7).
The model equation containing an additional non-
linear term that describes source-generated waves of
nite amplitude in a lossless uid is known as the
Riemann wave equation:
u
x


c
2
0
u
u

= 0 (8)
Physically, its general solution is u = f ( +ux/c
2
0
).
For sinusoidal source excitation, u = u
0
sin(t )
at x = 0, the solution is represented by the Fubini
series
1,2
:
u
u
0
=

n=1
2
nz
J
n
(nz) sin(n) (9)
Here z = x/x is the normalized coordinate (see
Fig. 3), and x = c
2
0
/(u
0
) is the shock formation
distance. As shown in Fig. 4, at the distance z =
1, or at x = x, the amplitude of the second and
third harmonics reach correspondingly 0.35 and 0.2
of the initial amplitude of the fundamental harmonic.
Consequently, at distances x x nonlinearity comes
into particular prominence.
For example, if an ultrasonic wave having an
intensity of 10 W/cm
2
and a frequency of 1 MHz
propagates in water ( 4), the shock formation
distance is about 25 cm. For a sound wave propagating
in the air ( 1.2) and having a sound pressure level
of 140 dB (relative to the root-mean-square pressure
162 FUNDAMENTALS OF ACOUSTICS AND NOISE
1
0.5
0 0.5
2u
2
2u
3
u
1
1
z
Figure 4 Schematic of energy pumping to higher
frequencies predicted by the Fubini solution.
2 10
5
Pa) and at a frequency 3.3 kHz, one can
estimate that x 6m.
Many of the physical phenomena accompanying
high-intensity wave propagation can compete with the
nonlinearity and weaken its effect. Phenomena such
as dissipation, diffraction, reection, and scattering
decrease the characteristic amplitude u
0
of the initial
wave and, consequently, increase the shock formation
distance x. The inuence of dissipation can be evalu-
ated by use of an inverse acoustical Reynolds number
= x,
1
where is the normal absorption coef-
cient of a linear wave. Numerical studies (Rudenko
23
)
show that nonlinearity is clearly observed at 0.1.
The absorption predominates at high values of , and
nonlinear transformation of the temporal prole and
spectrum is weak. For two examples given above, the
parameter is equal to 0.0057 (water) and 0.0014
(air), and conditions to observe nonlinear distortion
are very good.
The competition between nonlinearity and absorp-
tion is shown in Fig. 5. In the rst stage, for distances
x < x, the distortion of the initial harmonic wave pro-
le goes on in accordance with the Fubini solution [Eq.
(9)]. Thereafter, during the second stage, x < x < 2/,
a leading steep shock front forms inside each wave-
length, and the wave prole takes on a sawtooth-
shaped form. The nonlinear absorption leads to the
decay of the peak disturbance, and after considerable
energy loss has occurred, at distances x > 2/, the
wave prole becomes harmonic again. So, in this third
stage x > 2/, the propagation of the impaired wave
is described by the linear wave equation.
z = 0; 0.5; 1; 1.5;
3; 10; 20
0,5
0,5
1
0
1
V
q
p
p/2
p/2 p
= 0.1
Figure 5 Transformation of one period of harmonic initial signal in the nonlinear and dissipative medium.
23
Normalized
variables are used here: V = u/u
0
and Z = x/x.
NONLINEAR ACOUSTICS 163
3.2 General Wave Equation for Nonlinear
Wave Propagation
The general equation that describes one-dimensional
propagation of a nonlinear wave is
p
x
+
p
2
d
dx
ln S(x)

c
3
0

0
p
p

b
2c
3
0

2
p

2
= 0
(10)
Here p(x, ) is the acoustic pressure, which depends
on the distance x and the retarded time = t
x/c
0
, which is measured in the coordinate system
accompanying the wave and moves with the sound
velocity c
0
; , b are coefcients of nonlinearity and
effective viscosity,
1
and
0
is the equilibrium density
of the medium. Equation (1) can describe waves
traveling in horns or concentrators having a cross-
section area S(x). If S = const, Eq. (10) transforms
to the ordinary Burgers equation for plane waves.
1
If S x, Eq. (10) describes cylindrical waves, and if
S x
2
, it describes spherical ones. Equation (10) is
applicable also as a transfer equation to describe waves
propagating through media with large inhomogeneities
if a nonlinear geometrical approach is used; for such
problems, S(x) is the cross-section area of the ray
tube, and the distance x is measured along the central
curvilinear ray. Using new variables
V =
p
p
0
_
S(x)
S(0)
=
0
,
z =
0
p
0
x
_
0

c
3
0

0
_
S(0)
S(x

)
dx

one can reduce Eq. (10) to the generalized Burgers


equation
V
z
V
V

= (z)

2
V

2
(11)
whose properties are described in the literature.
1,17
Here p
0
and
0
are typical magnitudes of the initial
acoustic pressure and frequency, and
(z) =
_
b
0
2p
0
_
S(x)
S(0)
_
x=x(z)
(12)
is the normalized effective viscosity.
Next a one-dimensional model can be derived to
describe nonlinear waves in a hereditary medium (i.e.,
a medium with a memory)
1
:
p
x


c
3
0

0
p
p

m
2c
0

K(

)
p

= 0
(13)
Here m is a constant that characterizes the strength
of memory, and the kernel K(t ) is a decaying
function that describes the temporal weakening of
the memory. In relaxing uids K(t ) = exp(t /t
r
),
where t
r
is the relaxation time. Such an exponential
kernel is valid for atmospheric gases; it leads to the
appearance of dispersion and additional absorption,
which is responsible for shock front broadening during
the propagation of a sonic boom. For solids, reinforced
plastics and biological tissues K(t ) has a more
complicated form.
If it is necessary to describe the behavior of acous-
tical beams and to account for diffraction phenomena,
the following equation can be used:

(p)] =
c
0
2

p (14)
where

is the transverse Laplace operator acting


on coordinates in the cross section of the acousti-
cal beam;

(p) = 0 is one of the one-dimensional
equations that is to be generalized [e.g. Eqs. (10)
or (13)]. The KhokhlovZabolotskayaKuznetsov
Eq. (15)
5,9
:

_
p
x


c
3
0

0
p
p

b
2c
3
0

2
p

2
_
=
c
0
2
p (15)
is the most well-known example; it generalizes the
Burgers equation for beams and takes into account
diffraction, in addition to nonlinearity and absorption.
4 NONLINEAR TRANSFORMATION OF NOISE
SPECTRA
The following are examples and results obtained from
numerical or analytical solutions to the models listed
above. All tendencies described below have been
observed in laboratory experiments or in full-scale
measurements, for example, jet and rocket noise (see
details in the literature
1,14,16
).
4.1 Narrow-Band Noise
Initially, a randomly modulated quasi-harmonic signal
generates higher harmonics n
0
, where
0
is the fun-
damental frequency. At short distances in a Gaussian
noise eld the mean intensity of the nth harmonic is
n! times higher than the intensity of the nth harmonic
of a regular wave. This phenomenon is related to the
dominating inuence of high-intensity spikes caused
by nonlinear wave transformations. The characteristic
width of the spectral line of the nth harmonic increases
with increases in both the harmonic number n and the
distance of propagation x.
4.2 Broadband Noise
During the propagation of the initial broadband noise
(a segment of the temporal prole of the waveform
is shown by curve 1 in Fig. 6a) continuous distortion
occurs. Curves 2 and 3 are constructed for successively
164 FUNDAMENTALS OF ACOUSTICS AND NOISE
p
t
1
2
3
(a)
1
2
3
w
G(w, x)
(b)
Figure 6 (a) Nonlinear distortion of a segment of
the temporal prole of initial broadband noise p.
Curves 1, 2, and 3 correspond to increasing distances
x
1
= 0, x
2
> 0, x
3
> x
2
. (b) Nonlinear distortion of the
spectrum G(, x) of broadband noise. Curves 1, 2, and 3
correspond to the temporal proles shown in (a).
increasing distance and display two main tendencies.
The rst one is the steepening of the leading fronts and
the formation of shock waves; it produces a broadening
of the spectrum toward the high-frequency region.
The second tendency is a spreading out of the
shocks, collisions of pairs of them and their joining
together; these processes are similar to the adhesion of
absolutely inelastic particles and lead to energy ow
into the low-frequency region.
Nonlinear processes of energy redistribution are
shown in Fig. 6b. Curves 1, 2, and 3 in Fig. 6b are
the mean intensity spectra G(, x) of random noise
waves 1, 2, and 3 whose retarded time histories are
shown in Fig. 6a.
The general statistical solution of Eq. (10), which
describes the transformation of high-intensity noise
spectra in a nondissipative medium, is known for
b = 0
1
:
G(, x) =
exp
_

_

c
3
0

0
x
_
2
_
2
_

c
3
0

0
x
_
2

_
exp
_
_

c
3
0

0
x
_
2
R()
_
1
_
exp(i) d (16)
Here R( =
1

2
) = p(
1
)p(
2
) is the correlation
function of an initial stationary and Gaussian random
process, and
2
= R(0). For simplicity, the solution,
Eq. (16), is written here for plane waves; but one
can easily generalize it for arbitrary one-dimensional
waves [for any cross-section area S(x)] using the
transformation of variables. See Eq. (12).
4.3 NoiseSignal Interactions
The initial spectrum shown in Fig. 7a consists of
a spectral line of a pure tone harmonic signal and
broadband noise. The spectrum, after distortion by
nonlinear effects, is shown in Fig. 7b.
As a result of the interaction, the intensity of the
fundamental pure tone wave
0
is decreased, due to
the transfer of energy into the noise component and
because of the generation of the higher harmonics,
n
0
. New spectral wave noise components appear in
the vicinity = n
0
, where n = 1, 2, 3, . . . . These
noise components grow rapidly during the wave
propagation, ow together, and form the continuous
part of the spectrum (see Fig. 7b).
In addition to being intensied, the noise spectrum
can also be somewhat suppressed. To observe this
phenomenon, it is necessary to irradiate noise with an
intense signal whose frequency is high enough so that
the initial noise spectrum, and the noise component
generated near to the rst harmonic, do not overlap.
14
Weak high-frequency noise can be also partly sup-
pressed due to nonlinear modulation by high-intensity
low-frequency regular waves. Some possibilities for
the control of nonlinear intense noise are described in
the literature.
8,14
The attenuation of a weak harmonic signal due to
nonlinear interaction with a noise wave propagating in
the same direction occurs according to the law
p = p
0
exp
_

1
2
_

c
3
o

0
x
_
2
_
(17)
Signal Noise
G(w,0)
G(w, x)
w
w
3w
0
2w
0
w
0 0
(a)
(b)
Figure 7 Nonlinear interaction of spectra of the
tone signal and broadband noise. Initial spectrum
(a) corresponds to the distance x = 0. Spectrum
(b) measured at the distance x > 0 consists of higher
harmonics n
0
and new broadband spectral areas.
NONLINEAR ACOUSTICS 165
here
2
= p
2
is the mean noise intensity. The
dependence of the absorption with distance x in
Eq. (17) is given by exp(x
2
), which does not
depend on either the location of the noise or the signal
spectrum. The standard dependence exp(x) takes
place if a deterministic harmonic signal propagates in
a spatially isotropic noise eld. Here
21
=

2
4c
5
0

2
0
_
_

0
_
0

2
0
+
2

G() d +2
0

0
G() d
_
_
(18)
where G() is the spectrum of the noise intensity:

2
=

_
0
G() d
5 TRANSFORMATION OF STATISTICAL
DISTRIBUTION
The nonlinear distortion of the probability distribution
for nonlinear quasi-harmonic noise is illustrated in
Fig. 8. Curve 1 shows the initial Gaussian distribution.
Because of shock wave formation and subsequent
nonlinear absorption, the probability of small values of
the acoustic pressure p increases owing to the decrease
in the probability of large high-peak pressure jumps
(curves 2 and 3).
A regular signal passing through a random medium
gains statistical properties. Typical examples are con-
nected with underwater and atmospheric long-range
propagation, as well as with medical devices using
shock pulses and nonlinear ultrasound in such an inho-
mogeneous medium as the human body.
A sonic boom (N-wave) generated by a supersonic
aircraft propagates through the turbulent boundary
0
p
1
2
3
W(p)
Figure 8 Nonlinear distortion of the probability of
detection W(p) of the given value of acoustic pressure
p. Curves 1, 2, and 3 correspond to increasing distances
x
1
= 0, x
2
> 0, and x
3
> x
2
.
1
W(p)
2
3
0 p
0
p
Figure 9 Nonlinear distortion of the statistical distribu-
tion of the peak pressure of a sonic boom wave passed
through a turbulent layer. Line 1 is the intitial distribution,
curves 2 and 3 correspond to distances x
1
, x
2
> x
1
.
layers of the atmosphere. Transformation of the
statistical distribution of its peak pressure is shown
in Fig. 9.
18
Initial distribution is a delta-function (line
1), peak pressure is pre-determined and is equal to p
0
.
At increasing distances (after passing through the
turbulent layer, curves 2 and 3), this distribution
broadens; the probability increases that both small- and
large-amplitude outbursts are observed. So, turbulence
leads to a decrease in the mean peak pressure, but
uctuations increase as a result of random focusing
and defocusing caused by the random inhomogeneities
in the atmosphere.
Nonlinear propagation in media containing small
inhomogeneities responsible for wave scattering is
governed by an equation like Eq. (10), but one which
contains a fourth-order dissipative term instead of a
second-order one
19
:

b
2c
3
0

2
p

2
+

4
p

4
=
8
2
a
3
c
4
0
(19)
Here
2
is the mean square of uctuations of the
refractive index, and a is the radius of correlation.
Scattering losses are proportional to
4
instead of
2
in viscous media. Such dependence has an inuence on
the temporal prole and the spectrum of the nonlinear
wave; in particular, the increase of pressure at the
shock front has a nonmonotonic (oscillatory) character.
6 SAMPLE PRACTICAL CALCULATIONS
It is of interest in practice to consider the parameter
values for which the nonlinear phenomena discussed
above are physically signicant. For instance, in
measuring the exhaust noise of a commercial airliner
or of a spacecraft rocket engine at distances of 100 to
200 m, is it necessary to consider nonlinear spectral
distortion or not? To answer this question we evaluate
the shock formation distance for wave propagation in
air in more detail than in Section 3.
166 FUNDAMENTALS OF ACOUSTICS AND NOISE
For a plane simple harmonic wave, the shock
formation distance is equal to
x
pl
=
c
2
0
u
0
=
c
3
0

0
2fp
0
(20)
where p
0
is the amplitude of the sound pressure, and
f = /2 is the frequency. For a spherical wave
one can derive the shock formation distance using
Eqs. (10) and (11):
x
sph
= x
0
exp
_
c
3
0

0
2fp
0
x
0
_
(21)
Here x
0
is the radius of the initial spherical front of
the diverging wave. In other words, x
0
is the radius of
a spherical surface surrounding the source of intense
noise, at which the initial wave shape or spectrum is
measured.
Let the sound pressure level of the noise measured
at a distance of x
0
= 10 m be 140 dB, and the
typical peak frequency f of the spectrum be 1 kHz.
Evaluating the situation for propagation in air using
the parameters:
= 1.2 c
0
= 330 m/s
0
= 1.3 kg/m
3
gives the following values for the shock formation
distances:
x
pl
20 m x
sph
80m
So in this situation, shocks form in spherical wave
propagation at a greater distance than in plane wave
propagation because the spherical spreading decreases
the wave intensity and, consequently, nonlinear phe-
nomena accumulate more slowly in the spherical than
in the plane wave propagation case.
In all practical cases, the real shock front formation
length x obeys the inequality
x
pl
< x < x
sph
At distances, for which x x, nonlinear distortion is
signicant.
During experiments performed by Pestorius and
Blackstock,
25
which used a long tube lled with
air, strong nonlinear distortion of the noise spectrum
was observed at distances between as little as 2
to 10 m, for sound pressure levels of 160 dB. This
result agrees with predictions made using Eq. (20) for
a frequency f of about 1 kHz. Morfey
26
analyzed
several experiments and observed nonlinear distortion
in the spectra of four-engine jet aircraft at distances
between 262 and 501 m, for frequencies between f =
2 and 10 kHz. He also analyzed the noise spectrum
of an Atlas-D rocket engine at distances of 1250
to 5257 m, at frequencies in the range of f = 0.3
to 2.4 kHz. These observations correspond to the
analytical case of spherically diverging waves. See
Eq. (21).
Extremely strong noise is produced near the
rocket exhausts of large spacecraft during launch.
For example, assume that the sound pressure level is
170 dB at 10 m from a powerful space vehicle such as
the Saturn V or the space shuttle. The shock formation
distance is predicted from Eq. (21) to be a further
distance of x
sph
13 m for a frequency of 500 Hz.
The approximate temporal duration t
fr
of the shock
front at a distance x can be calculated using Eq. (22),
which is found in Rudenko and Soluyan
1
:
t
fr
=
1

2
f
x
pl
x
abs
x
x
0
_
1 +
x
0
x
pl
ln
_
x
x
0
__
(22)
where x
abs
=
1
= (4
2
f
2
)
1
is the absorption
length, and the value of = 0.5 10
12
s
2
/m is
assumed for air. For the assumed sound pressure level
of 170 dB and the frequency 500 Hz, we substitute
the values x
pl
= x
0
= 10 m and evaluate the width
of the shock front l
fr
= c
0
t
fr
at small distances of
25 to 30 m from the center of the rocket exhaust
nozzles. This shock width being of the order of l
fr

0.01 0.1 mm is much less than the wavelength,
67 cm. Such a steep shock front is formed
because of strong nonlinear effects. As the sound wave
propagates, the shock width increases and reaches
l
fr
7 cm at distances of about 23 km. It is evident
that nonlinear phenomena will be experienced at large
distances from the rocket. That is the reason why it
is possible to hear the crackling sound standing far
from the launch position. However, the value of 23
km for the distance at which the shock disappears
is realistic only if the atmosphere is assumed to be
an unlimited and homogeneous medium. In reality,
due to reection from the ground and the refraction
of sound rays in the real inhomogeneous atmosphere,
the audibility range for shocks can be somewhat less.
To describe nonlinear sound propagation in the real
atmosphere, the numerical solution of the analysis
of more complicated mathematical models such as
Rudenko
22
needs to be undertaken.
It is necessary to draw attention to the strong
exponential dependence of nonlinear effects on the
frequency f
0
, the sound wave amplitude p
0
, and the
initial propagation radius x
0
, for spherical waves. From
Eq. (21) we have
x
sph
= x
0
exp
_
const
fp
0
x
0
_
Consequently, the shock formation distance x
sph
is
very sensitive to the accuracy of measurement of these
parameters. Other numerical examples concerning
nonlinear noise control are given in the literature.
8,14,16
Consider now a sonic boom wave propagating as a
cylindrical diverging wave from a supersonic aircraft.
NONLINEAR ACOUSTICS 167
The shock formation distance for this case is
x
cyl
= x
0
_
1 +
c
3
0

0
4fp
0
x
0
_
2
(23)
At small distances from the aircraft, for example,
at 50 m, the peak sound pressure is about 3000 Pa
and the pulse duration t
0
= f
1
l/v, where l is the
length of the aircraft fuselage, and v > c
0
is the speed
of supersonic ight. For the parameters of aircraft
length and speed, l = 10 m, v = 1.3c
0
, evaluation of
Eq. (23) gives x
cyl
100 m. This means that at several
hundred metres from the aircraft, the multiple colli-
sions of shocks generated by singularities of the aero-
dynamic prole come to an end, and the sonic boom
wave changes into an N-wave, as shown in Fig. 2b.
At greater distances, the peak pressure of the N-
wave decreases, and the value of the distance x
cyl
increases. For a peak pressure of 200 Pa measured at
x
0
= 1 km, and for a pulse duration t
0
= f
1
= 0.05s
we obtain a distance of x
cyl
3 km, according to Eq.
(23). So, for a distance of x
0
= 1 km from the aircraft,
an additional distance x
cyl
x
0
= 2 km will produce
a signicant change in the shape of the N-wave due to
the nonlinear wave propagation effects.
Nonlinear phenomena appear also near to the sharp
tips of bodies and orices in the high-speed streamlines
of an oscillating uid. These nonlinearities are caused
by the large spatial gradients in the hydrodynamic eld
and are related to the convective term (u) u in the
equation of motion of the uid, in the form of the
NavierStokes or Euler equations. This effect is quite
distinct from the more common nonlinear phenomena
already described. Nonlinear wave distortion cannot
build up during wave propagation since the effect only
has a local character. To determine the necessary
sound pressure level at which we can observe these
phenomena in an oscillating ow, we evaluate the
velocity gradients. (Note that in the case of harmonic
vibrations in the streamlines around an incompressible
liquid, higher harmonics will appear.) We assume that
the gradient is of the order of u/ max(r
0
, ), where =

/ is the width of the acoustical boundary layer,


r
0
is the minimum radius of the edge of the body, u is
the vibration velocity, and = /
0
is the kinematic
viscosity. The width is the dominating factor for sharp
edges, if r
0
< . This boundary nonlinearity is sig-
nicant at Reynolds numbers of Re 1, which are pro-
portional to the ratio of the terms in the NavierStokes
or Euler equations of motion of the uid:
Re

(u)u
_
u
t
_
1

=
_
2I
c
(24)
As can be determined by Eq. (24), this nonlinearity
manifests itself in air at a sound pressure level of
120 dB, at a frequency of about 500 Hz. If vortices
form near to the edge of a body immersed in an
oscillating ow, nonlinearity in such a ow can be
observed even at a sound pressure level as low
as 90 dB. Boundary layer nonlinearity is signicant
in the determination of the resonance frequency of
sound absorbers, which contain Helmholtz resonators
with sound-absorbing material in their necks. This
nonlinearity can detune the resonance condition at the
frequency given by linear approximations. It can even
have the opposite effect of enhancing the dissipation
of acoustic energy by the absorber, if it is excited off
resonance, according to the linear approximations.
8
7 FURTHER COMMENTS AND CONCLUSIONS
Only common nonlinear events occurring in typical
media have been discussed. However, nonlinear phe-
nomena of much more variety can occur. Nonlinearity
manifests itself markedly in conditions of resonance, if
the standing waves that form in spatially limited sys-
tems have a high Q factor. Using high-Q resonators,
it is possible to accumulate a considerable amount of
acoustic energy and provide conditions for the clear
manifestation of nonlinear phenomena even in the case
of weak sound sources.
23
Some structures (such as components of the fuse-
lage of an aircraft) can have huge nonlinearities caused
by special types of inhomogeneous inclusions (in
cases such as the delamination of layered compos-
ites, with cracks and grain boundaries in metals, and
with clamped or impacting parts, etc.). These nonlin-
ear phenomena can be used to advantage in sensitive
nondestructive tests.
It is necessary to mention a nonlinear device
known as a parametric array. Its use is common in
underwater acoustics.
9
Recently, it has also been put to
use in air in the design of parametric loudspeakers.
27,28
The difference between linear and nonlinear prob-
lems is sometimes only relative. For example, aero-
dynamic sound generation can be referred to as a
linear problem; but some people say that this is a non-
linear phenomenon described by the nonlinear terms
in the Lighthill equation. Both viewpoints are true.
Chapter 9 in this book, which is written by Morris
and Lilley, is devoted to the subject of aerodynamic
sound. The aerodynamic exhaust noise generated by
turbojet and turbofan engines is discussed by Huff and
Envia in Chapter 89 of this book. Lighthill, Powell
and Ffowcs Williams also discuss jet noise genera-
tion in Chapters 24, 25, and 26 in the Handbook of
Acoustics.
29
REFERENCES
1. O. V. Rudenko and S. I. Soluyan, Theoretical Foun-
dations of Nonlinear Acoustics, Plenum, Consultants
Bureau, New York, 1977.
2. R. T. Beyer, Nonlinear Acoustics in Fluids, Van Nos-
trand Reinhold, New York, 1984; Nonlinear Acoustics,
Acoustical Society of America, American Institute of
Physics, New York, 1997.
3. K. A. Naugolnykh and L. A. Ostrovsky (Eds.), Non-
Linear Trends in Physics, American Institute of Physics,
New York, 1994.
4. K. Naugolnykh and L. Ostrovsky, Non-Linear Wave
Processes in Physics, Cambridge University Press,
Cambridge, 1998.
168 FUNDAMENTALS OF ACOUSTICS AND NOISE
5. M. F. Hamilton and D. T. Blackstock, Nonlinear
Acoustics, Academic, San Diego, 1998.
6. O. V. Rudenko, Nonlinear Acoustics, in Formulas of
Acoustics, F. P. Mechel (Ed), Springer, 2002.
7. L. Kinsler, Frey et al., Fundamentals of Acoustics, 4th
ed., Wiley, New York, 2000.
8. O. V. Rudenko and S. A. Rybak (Eds.), Noise Control
in Russia, NPK Informatica, 1993.
9. B. K. Novikov, O. V. Rudenko, and V. I. Timoshenko,
Nonlinear Underwater Acoustics (trans. R. T. Beyer),
American Institute of Physics, New York, 1987.
10. G. C. Stokes, On a Difculty in the Theory of Sound,
Philosophical Magazine, Ser. 3, Vol. 33, 1848, pp.
349356.
11. D. T. Blackstock and M. J. Crocker, in Handbook of
Acoustics, M. J. Crocker (Ed.), Wiley, New York, 1998,
Chapter 15.
12. D. T. Blackstock, in Handbook of Acoustics, M. J.
Crocker (Ed.), Wiley, New York, 1998, Chapter 16.
13. D. G. Crighton, in Handbook of Acoustics, M. J.
Crocker (Ed.), Wiley, New York, 1998, Chapter 17.
14. O. V. Rudenko, Interactions of Intense Noise Waves,
Sov. Phys. Uspekhi, Vol. 29, No.7, 1986, pp. 620641.
15. S. N. Gurbatov, A. N. Malakhov, and A. I. Saichev,
Nonlinear Random Waves and Turbulence in Nondis-
persive Media: Waves, Rays, Particles, Wiley, New
York, 1992.
16. S. N. Gurbatov, and O. V. Rudenko, Statistical Phe-
nomena, in Nonlinear Acoustics, M. F. Hamilton and
D. T. Blackstock (Eds.), Academics, New York, 1998,
pp. 377398.
17. B. O. Eno and O. V. Rudenko, To the Theory of
Generalized Burgers Equations, AcusticaActa Acus-
tica, Vol. 88, 2002, pp. 155162.
18. O. V. Rudenko and B.O. Eno, Nonlinear N-wave
Propagation through a One-dimensional Phase Screen,
AcusticaActa Acustica, Vol. 86, 2000, pp. 229238.
19. O. V. Rudenko and V. A. Robsman, Equation of
Nonlinear Waves in a Scattering Medium, Doklady-
Physics (Reports of Russian Academy of Sciences), Vol.
47, No. 6, 2002, pp. 443446.
20. D. T. Blackstock, Nonlinear Acoustics (Theoretical),
in American Institute of Physics Handbook, 3rd ed.
D. E. Gray (Ed.), McGraw-Hill, New York, 1972, pp.
3-1833-205.
21. P. J. Westervelt, Absorption of Sound by Sound,
J. Acoust. Soc. Am., Vol. 59, 1976, pp. 760764.
22. O. V. Rudenko, Nonlinear Sawtooth-shaped Waves,
PhysicsUspekhi, Vol. 38, No 9, 1995, pp. 965989.
23. O. A. Vasileva, A. A. Karabutov, E. A. Lapshin, and
O. V. Rudenko, Interaction of One-Dimensional Waves
in Nondispersive Media, Moscow State University
Press, Moscow, 1983.
24. B. O. Eno, C. M. Hedberg, and O. V. Rudenko,
Resonant Properties of a Nonlinear Dissipative Layer
Excited by a Vibrating Boundary: Q-factor and Fre-
quency Response, J. Acoust. Soc. Am., Vol. 117, No. 2,
2005, pp. 601612.
25. F. M. Pestorius and D. T. Blackstock, in Finite-
Amplitude Wave Effects in Fluids, IPC Science and
Technology Press, London, 1974, p. 24.
26. C. L. Morfey, in Proc. 10 International Symposium on
Nonlinear Acoustics, Kobe, Japan, 1984, p. 199.
27. T. Kite, J. T. Post, and M. F. Hamilton, Parametric
Array in Air: Distortion Reduction by Preprocessing,
in Proc. 16th International Congress on Acoustics, Vol.
2, P. K. Kuhl and L. A. Crum (Eds.), New York,
ASA, 1998, pp. 10911092.
28. F. J. Pompei, The Use of Airborne Ultrasonics for
Generating Audible Sound Beams, J. Audio Eng. Soc.,
Vol. 47, No. 9, 1999.
29. M. J. Crocker (Ed.), Handbook of Acoustics, Wiley, and
New York, 1998, Chapters 23, 24, and 25.

You might also like