ail

*-

I Propeller and I

Propfan Noise

Lead author
B. Magliozzi
Hamilton Standard Division-
United Technologies
Windsor Locks, Connecticut
Contributing authors
D. B. Hanson R. K. Amiet
Hamilton Standard Division- AMI
United Technologies Wooster, Ohio
Windsor Locks, Connecticut

d3

Introduction
Propellers are familiar devices. Indeed, these were the first means of powering
airplanes, preceding all other means of propulsion by about 40 years. Propellers were
used extensively through the 1940’s. Although there have been many refinements to
propellers through the years, such as variable pitch and the application of composite
materials to reduce weight, the general appearance of the propeller has changed little.
A propeller can be generally described as an open (unshrouded), rotating, bladed
device. Although there are many differences in details among various designs
and applications, such as number of blades, blade shape, and airfoil section, the
noise-generating process is basically the same for all. The major propeller noise
components are thickness noise (due to the volume displacement of the blades),
steady-loading noise (due to the steady forces on the blades), unsteady-loading noise
(due to circumferentially nonuniform loading), quadrupole (nonlinear) noise, and
broadband noise. Although the relative importance of these sources depends on
design and operating conditions, defining them will completely describe the acoustic
signature of a propeller.
One important consideration is the effect of installation on the noise produced
by a propeller. This effect is essentially the difference between the laboratory
environment and the real world. It is generally assumed that in a laboratory
environment conditions are ideal, that is, the propeller is operating in perfectly
uniform flow. For an operational propeller, this is never the case. Propellers are
always operating in a flow field that has some distortion. This can be from the wing
upwash, the pylon wake, the airplane angle of attack, or the inflow turbulence. Since
this distortion leads to additional noise, it is a factor which must be considered in
defining the total noise of an operational propeller.
Magliozzi, Hanson, and Amiet

In this chapter the characteristics of propellers, their noise-generating mecha-

nisms, propeller noise theories and calculation procedures, sound propagation effects,
comparisons of predictions and measurements, and means for controlling propeller
noise are described.

Description of Propellers

General Characteristics
A propeller is an open rotor having fixed or adjustable-pitch blades. The blades
are designed to produce a region of low pressure on one side and high pressure
on the other. The resulting forces induce air from the front and push it back,
resulting in thrust. Because propellers impart a relatively small amount of velocity
to a large mass of air, their efficiencies are high. However, the efficiencies of
conventional propellers tend to fall off at high speed. This has led to a variation
of the propeller called the propfan. The propfan is also an open rotor, but compared
with conventional propellers it has a smaller diameter for a given thrust and has
more blades, which are swept. To improve efficiency further, a second rotor can be
added behind a propeller, resulting in a counterrotating propeller.
A typical high-performance, modern propeller in widespread use on commuter
airplanes is shown in figure 1. These propellers have two to six relatively straight,
narrow blades. Although this type of propeller is well-suited for moderate flight
speed (below a Mach number of about 0.65), its performance decreases at higher
flight speeds. The primary limitation of this propeller is associated with high drag
at high speed due to blade thickness and large relative blade section speed.
The propfan, shown in figure 2, has been developed to extend the practical flight
envelope of the propeller. Compared with conventional propellers, the propfan has
more blades (from 6 to 12), uses thin airfoils and swept blades to provide good
aerodynamic performance at high speed, and operates at much higher power loading
to reduce the diameter.
One factor leading to loss of efficiency in propellers is swirl in the wake resulting
from engine torque. Generation of swirl uses energy but does not contribute to
thrust. The amount of swirl is related to the power loading. One concept to recover
the swirl losses is to add a second blade row behind the first. This is shown in figure 3
for the propfan. The second blade row rotates in the direction opposite to that of the
first, thus cancelling its swirl. This cancellation can result in performance increases
of 8 to 10 percent compared with that of single rotation propellers (ref. 1).

Installation of Propellers and Propfans

In considering the noise of propellers and propfans, it is important to address
the installation of these devices, as this can have a significant effect on the noise-
generation process. In their simplest forms noise calculation procedures and analyses
assume uniform conditions, that is, the loads on the blades are absolutely steady.
In actuality, that is rarely the case. Although laboratory tests can be conducted
such that the incoming flow is uniform and free of turbulence, the real environment
is never as ideal. The amount of distortion is generally related to two parameters:
operating conditions and installation.

2
 Propeller and Propfan Noise

Figure P. High-performance, dow-speed Figwe 2. Propfun propulsion

propellerfor moderaie-flight-speedwm- system for high subsonic
muter aiplane propulsion. cruise speed applacations.

Figure 3. Co2llaierrotating-pP~propulsion system for high subsonic cm&e

speed applicutioas.

3
Magliozzi, Hanson, and Amiet

Operating conditions include static (zero flight speed) or forward flight. In flight,
the propeller can be at an angle of attack. Static conditions give rise to severe inflow
distortion and the resulting noise is vastly different than that from propellers in
flight. This difference is discussed in more detail in a subsequent section. Angle
of attack or skewed inflow causes unsteady loading, with periodicities equal to one
cycle per rotation.
hgarding installations, the cleanest configuration is in front of a long, slender,
axisymmetric nacelle. The worst is probably behind a wing. Intermediate configu-
rations include wing-mounted tractors, aft-mounted tractors in front of a pylon, and
aft-mounted pushers behind a pylon. These installations result in varying degrees of
inflow distortion which typically results in added sources (unsteady loading noise)
and increases the noise produced by the propeller.
It is thus important to evaluate the propeller as an installed system rather than
as an isolated component when noise requirements are addressed. If a propeller is
designed to meet the noise goals, even with a comfortable margin of error, ignoring
installation effects can result in a substantial underprediction of the system noise,
with the strong possibility that the airplane will not meet the noise requirements.

Propeller Noise Characteristics

Propeller noise can be classified into three categories: haxmonic noise, broadband
noise, and narrow-band random noise.
Harmonic noise is the periodic component, that is, its time signature can be
represented by a pulse which repeats at a constant rate. If an ideal propeller with B
blades is operating at constant rotational speed N , then the resulting noise appears
as a signal with fundamental frequency B N . The blade-passage period is 1/BN.
Typically the generated pulse is not a pure sinusoid, so that many harmonics exist.
These occur at integer multiples of the fundamental frequency. The first harmonic
is the fundamental, the second harmonic occurs at twice the fundamental frequency,
and so on. Figure 4 illustrates the characteristics of harmonic noise in both the time
and frequency domains.
Broadband noise is random in nature and contains components at all frequencies.
A typical broadband noise signal for propellers is shown in figure 5. The frequency
spectrum is continuous, although there may be a “shape” to it because not all
frequencies have the same amplitude.
Narrow-band random noise is almost periodic. However, examination of the
harmonics reveals that the energy is not concentrated at isolated frequencies, but
rather it is spread out. As illustrated in figure 6 the signal may appear periodic, but
certain components do not repeat exactly with time. The frequency spectrum shows
discrete components, but these spread out, particularly at the higher frequencies.

Propeller Noise-Generating Mechanisms

The mechanisms which lead to the generation of the spectral characteristics

discussed above are described in this section.

4
 Propeller and Propfan Noise

Amplitude

t-4
-
\

Fundamental frequency = BN

BN 2BN
Frequency
3BN
I 4BN

Figure 4. Characteristics of propeller rotational noise.

Steady Sources

Steady sources are those which would appear constant in time to an observer
on the rotating blade. They produce periodic noise because of their rotation. Noise
sources are usually divided into three categories: linear thickness, linear loading, and
(nonlinear) quadrupole.
Thickness noise arises from the transverse periodic displacement of the air by
the volume of a passing blade element. The amplitude of this noise component is
proportional to the blade volume, with frequency characteristics dependent on the
shape of the blade cross section (airfoil shape) and rotational speed. Thickness noise
can be represented by a monopole source distribution and becomes important at
high speeds. Thin blade sections and planform sweep are used to control this noise.

5
Magliozzi, Hanson, and Amiet

I >
Time

: Frequency

Figure 5. Characteristics of propeller broadband noise.

Loading noise is a combination of thrust and torque (or lift and drag) components
which result from the pressure field that surrounds each blade as a consequence of
its motion. This pressure disturbance moving in the medium propagates as noise.
Loading is an important mechanism at low to moderate speeds.
For moderate blade section speed, the thickness and loading sources are linear and
act on the blade surfaces. When flow over the blade sections is transonic, nonlinear
effects can become significant. In aeroacoustic theory these can be modeled with
quadrupole sources distributed in the volume surrounding the blades.
In principle, the quadrupole could be used to account for all the viscous and
propagation effects not covered by the thickness and loading sources. However, the
only practical application of this term to propeller acoustics has been its evaluation
in the nonviscous flow close to the blade surface. At transonic blade section speeds
the quadrupole enhances the linear thickness and loading sources and causes a noise
increase for unswept, high-tip-speed propellers and helicopter rotors.

6
 Propeller and Propfan Noise

Amplitude

Time
>

Spectrum
level

I I I I I >
BN 2BN 3BN 4BN
Frequency

Figure 6. Characteristics of propeller narrow-band random noise.

Unsteady Sources

Unsteady sources are time dependent in the rotating-blade frame of reference.

They include periodic and random variation of loading on the blades.
A typical example of periodic blade loading in propellers is the effect of shaft
angle of attack. When the propeller axis is tilted relative to the inflow, each blade
sees a cyclic change in local angle of attack. As a consequence, the loading on the
blade varies during a revolution. The loading change may be once per revolution or
several times per revolution, depending on the source of inflow distortion. All inflow
distortion which is invariant with time results in blade-loading changes which repeat

7
Magliozzi, Hanson, and Amiet

exactly for every propeller revolution. The resulting periodic unsteady-loading noise
occurs at harmonics of blade-passage frequency.
Depending on the circumferential location of the loading disturbance relative to
the observer, unsteady-loading noise can add or subtract from the steady-loading
noise. The noise directivity is no longer axisymmetric and a third coordinate
is needed to define it. The circumferential directivity exhibits lobes-peaks and
valleys-with the number of lobes dependent on the order of the distortion and
unrelated to the number of blades. For example, a propeller behind a wing might
show two circumferential directivity lobes regardless of the number of blades on the
propeller.
Unsteady loading is an important source in the counterrotating propeller. Al-
though the counterrotating propeller does not contain any additional or unique
sources of noise, the aerodynamic interference between the two rotors gives rise to
significant levels of unsteady-loading noise which are particularly significant at low
flight speeds, such as during takeoff and landing. Each front rotor blade leaves a
wake which convects into the rear rotor. (This wake can be complex, consisting of
downwash due to the lift on the blades, velocity deficits due to the drag of the blade
sections, and tip vortices.) This convection results in a sequence of lift pulses on the
rear rotor blades. Another mechanism is the potential field (due to blade loading) of
the rear rotor creating a disturbance which is felt by the aft part of the front rotor
blades. The magnitude of this source depends on the level of loading on the rear
rotor and the spacing between the two rotors.
Because the wakes are periodic, the generated noise is also periodic. If the two
rotors have the same number of blades and are operating at the same rotational
speed, then the components of the steady sources and the unsteady sources are at
the same frequencies and the noise spectra contain only harmonics of blade-passage
frequency. However, if the number of blades of the front rotor is different from that
of the rear rotor or the two rotors operate at different rotational rates, then the
individual interaction components (modes) are distinct in the noise spectra.
Figure 7 shows the importance of the aerodynamic interaction in a counter-
rotating propeller (ref. 2). Figure 7(a) shows the spectra of single-rotating-propeller
(SRP) noise at a forward directivity, near the plane of rotation, and at an aft
directivity. It is readily apparent that there are no significant higher frequency
harmonics. For comparison, noise spectra for a counterrotating propeller (CRP)
are shown in figure 7(b). It is apparent that the counterrotating propeller has
significantly higher levels of higher frequency harmonics. Figure 7(c) shows a direct
comparison between the noise from the two types of propellers. In this comparison
the two propellers were operating at equal tip speeds and power per rotor. Three
decibels were added to the SRP noise levels to simulate the total power of the
counterrotating propeller. If the two rotors of the counterrotating propeller were
uncoupled, then the two spectra would be identical. In fact, the levels at the
blade-passage frequency are very close. At the higher harmonics, the counter-
rotating propeller shows significantly higher levels. This is a direct indication of
the aerodynamic interaction effects on noise in counterrotating propellers.
Aerodynamic interaction is a significant source of noise for low-speed opera-
tion. At higher flight speeds, such as during cruise, the aerodynamic interaction
becomes less important because the steady sources (thickness, steady loading, and
quadrupole) become dominant.

8
 Propeller and Propfan Noise

Plane of rotation
120 -

110-

0 1 2 3 0 1 2 3 0 1 2 3
Frequency, kHz
(a) Single-rotating-propeller noise spectra.

120
lane of rotation
110

100

SPL, dB 90

80

70

60
0 1 2 3 0 1 2 3 0 1 2 3
Frequency, kHz

(b) Counterrotating-propeller noise spectra.

Figure 7. Aerodynamic interaction noise in counterrotating propeller. (From

ref. 2.)

9
Magliozzi, Hanson, and Amiet

Forward Plane of rotation Aft

120 [Interaction t
110 -

100 -

SPL, dB
90 -

*O

70
t-SRP+3dB 2
,
noise

,
1 2 3 4 5
, , I ,

1 2 3 4 5
Harmonic of blade-passage frequeiicy
, , I

(c) Aerodynamic interaction noise.

Figure 7. Concluded.

Under certain conditions blade loading which is nearly periodic can occur. An
example of this is the ingestion of a vortex, which could be induced by the propeller
and attached to the fuselage or to the ground ahead of the propeller. In this example,
a local distortion is induced by and drawn into the propeller. The blades chop
through the distortion and a blade-loading pulse is produced. Because the distortion
can persist for several propeller revolutions, the unsteady-loading noise can appear
at blade-passage frequency harmonics. However, as conditions change, the location
of the distortion changes and the amplitude and phase of the unsteady-loading noise
change. These changes broaden the noise spectrum, as previously described for
narrow-band random noise.

Random Sources
Random sources give rise to broadband noise. For propellers there are two sources
which may be important, depending on the propeller design and operating conditions.
The first broadband noise source is the interaction of inflow turbulence with
the blade leading edges. Because the inflow is turbulent, the resulting noise is
random. The importance of this noise source depends on the magnitude of the
inflow turbulence, but it can be quite significant under conditions of high turbulence
at low speeds.
In the second broadband mechanism, noise is generated near the blade trailing
edge. A typical propeller develops a turbulent boundary layer over the blade surfaces,
which can result in fluctuating blade loading at the trailing edge. The noise is
characterized by the boundary-layer properties. A related mechanism occurs at the
blade tips, where turbulence in the core of the tip vortex interacts with the trailing
edge.

10
 PropeEEer and Propfan Noise

It has been determined for full-scale propellers in flight that the broadband noise
sources are relatively unimportant and do not contribute significantly to the total
noise (ref. 3).

Prediction Methods for Propeller

Harmonic Noise
Any of the propeller noise methods currently in use can be derived from the
Ffowcs Williams and Hawkings equation (ref. 4). This fundamental equation of
sound generation is attractive because it is a rigorous combination of the equations
of momentum, continuity, and state into a wave equation that can be solved to
varying degrees of precision by a variety of analytical methods. A scholarly survey
of these methods was presented by Farassat in 1981 (ref. 5 ) and is recommended for
readers interested in the mathematical connections between these methods and the
Ffowcs Williams-Hawkings equation.
In the following sections, the noise radiation equations which are discussed
were chosen to illustrate the variety of methods available, and comments on their
advantages and disadvantages are included. In the case of the time-domain equations,
computer programs based on the theory of Farassat are available to U.S. companies
from NASA Langley Research Center. The frequency-domain equations are simple
enough so that they can be coded on personal computers. Furthermore, their
analytical form gives a direct indication of the influences of propeller design features
on noise characteristics.

Linear Theories
As given by Farassat (ref. 5 ) , the linear form of the Ffowcs Williams and Hawkings
equation is

where the left side is the well-known linear wave operator acting on the acoustic
pressure p . The right side contains the source terms resulting from the motion of
surfaces in the fluid: po is the ambient density, c is the ambient speed of sound, v,
is the local velocity of the surface normal to itself, S(f) is the Dirac delta function,
zi is the observer position, and Zi is the ith component of the surface force. The first
source term represents the effect of the blades parting the air and produces what
is known as “thickness noise.” The second term represents the action of the blade
forces on the air and produces “loading noise.”
In equation (l), the presence of the surfaces is accounted for by the factors
containing f, where f = 0 is the equation of the blade surface. Unless very high
frequencies are considered (wavelengths of the order of blade thickness), details of the
airfoil section can be ignored. The source term is thus simplified, so that equation (1)
becomes
v p2 - - - =1-a2P
p - q+ v . j 7
a
c2 at2 O at

where now the thickness source can be thought of as being represented by a volume
distribution af sources (and sinks) of strength q. The loading source is represented by

11
Magliozzi, Hanson, and Amiet

a volume distribution of doublets associated with F, the force imposed by the blades
on the air. Equation (2) is the linear wave equation with sources as derived, for
example, in chapter 7 of reference 6 and in chapter 1 of reference 7. This source and
doublet concept is equivalent to the common representation of wing aerodynamics in
textbooks. For example, the sketch in figure 8, adapted from reference 8, was used
to treat the thickness and loading effects of wings. The dominant sources are exactly
the same for wing and propeller aerodynamics. However, in the case of propellers,
the sources make noise because of unsteadiness associated with rotation and with
time-dependent blade loading.

Blade or
wing at angle - Thickness + +
Camber incidence
of attack effect effects

Source
representation - Sources Sinks Vortices

Figure 8. Decomposition of wing or blade section aerodynamics into

thickness effect and camber and incidence eSfects. (Based on ref. 8.)

Time-domain methods are used to solve equations (1) or (2) directly in terms
of the space-time variables. These methods are appealing because they can treat
blade geometry with any desired level of precision. The result is the prediction of
the acoustic pressure waveform p ( t ) . If noise harmonics are needed, p ( t ) is Fourier
transformed numerically.
Frequency-domain methods eliminate time from the wave equation by means of
Fourier transformation. Some precision in the representation of blade geometry
is usually lost through the transformation, but this loss is generally acceptable
for harmonics to a fairly high order. The transformation also gives rise to Bessel
functions which are indicators of radiation efficiency. Harmonics are computed one
at a time and waveforms are generated by summing a Fourier series.

Time-Domain Methods
The most prolific proponent of time-domain methods for propellers and rotors
has been Farassat. Papers listed in the References section can be used to trace the

12
 Propeller and Propfan Noise

development of his formulations 1, lA, 2, and 3. The ed formulations are

coded in the Aircraft Noise Prediction Program (ANOPP, ref. 9) and the Dunn-
Farassat-Padula Advanced Turboprop Prediction (DFP-ATP, ref. 10) program and
are described briefly below.
Formulation lA, used in both ANOPP and DFP-ATP for subsonic source regions,
gives the acoustic pressure p ( t ) as follows:

The dots on Mi (defined as wi/c) and li denote derivatives with respect to source
time, and wi is the local velocity of the blade surface with respect to the quiescent
fluid. The first three integrals give the loading noise, with the blade load vector li
as the source. The last integral gives the thickness noise, with the surface-normal
velocity v, as the source. Also, r is the distance from a source point to the observer
and M, is the Mach number of the source toward the observer. Formulation 3
(not shown herein) is recommended by Farassat for supersonic blade sections and
is coded in DFP-ATP (ref. 10). It too contains integrals in the form &, [ Iret dS.
The significance of this notation is that the contents of the brackets are evaluated
on the surface f = 0 at the retarded (or emission) time and integrated over the
blade surface elements dS. Thus, to compute the acoustic pressure at time t , it must
be determined where every element of the surface was when it emitted the wavelet
that arrives at the observer point at t. This is possible because the geometry and
motion of the blades are known and it is assumed that the wavelets travel at the
ambient speed of sound. When a surface is constructed by connecting all the blade
edges at their retarded times, the result is distorted from the physical planform to a
shape called the “acoustic planform.” The process is illustrated by simple example
in figure 9 (from a paper by Hanson, ref. ll), which also provides the key to the
acoustic planforms shown in figure 10. The blade labeled “visual planform” has a
rectangular shape and rotates in a plane with zero forward flight speed. At a tip
Mach number Mt of 1.1,a sharp thickness noise pulse like the one at the right results
in accordance with formulation 3 for an observer at a distance of 5 rotor diameters.
The acoustic planform shown on the source disk is for the current time indicated by
the dot on the waveform. Figure 10 shows the evolution of the acoustic planform
with time. For this supersonic example, the planform breaks into two portions
because, for some blade elements, there is more than one point on the azimuth
where waves are emitted that arrive at time t. For subsonic tip speeds, the acoustic
planform is in one continuous piece and is straightforward to calculate. However,
the problem at supersonic speeds is surprisingly difficult and leads to significant
numerical problems because the acoustic planform must be evaluated with great
precision for good results.
As mentioned above, computer programs embodying Farassat’s formulas are
available outside NASA on a limited basis for personal and mainframe computers.
For users with no desire to do their own coding, this offers a means to acquire
noise prediction capability quickly. Because of the numerical difficulties associated

13
Magliozzi, Hanson, and Amiet

+
I Observer

corresponding
to planform
plot
u
Source
disk
Visual planform
Mt = 1.1

Figure 9. Key to acoustic planform plots. (From ref. 11.)

t ; Time

-!-
t +
Figure 10. Evolution of acoustic planform with time at Mt = 1.1. (From ref. 11.)

14
 Propeller and Propfan Noise

with the acoustic planform determination and numerical differentiation, coding of

equation (3) should not be attempted without careful study of references 5, 9, 12,
13, and 14.

fiequency-Domain Methods
A transformation to the frequency domain eliminates the need for computing
retarded blade locations and numerical derivatives. By representing blades as
helicoidal surfaces, far-field noise formulas that are easily coded on a personal
computer can be derived. Effective radius versions can even be computed by hand
with the help of a Bessel function table. Furthermore, these formulas give direct
insight to the influence of blade geometry and operating conditions on the sound
harmonics.
The first successful propeller noise theory by Gutin (ref. 15) was in harmonic
form. This theory was extended by various investigators; one of these was Hanson,
whose versions included effects of thickness, forward flight, and blade sweep (refs. 16
to 18). Hanson’s formulas are described below in enough detail for programming. To
arrive at his simple results, the approximation is made that the thickness and loading
sources act on the advance helix, that is, on the surface swept out by a radial line
that rotates at angular speed 0 and translates at the flight speed V. Of course, the
aerodynamic loading comes from the fact that blades are at an angle of attack relative
to the helical surface. However, once the loading is determined from an aerodynamic
analysis, the thickness and loading sources are transferred to the advance helix for
the radiation calculation. This transfer corresponds to linearization of the boundary
condition to the free-stream direction in wing theory. With this simplification, the
sources can be modeled with the terms on the right in equation (2), and the far-field
pressure can be found from the free-space Green’s function in the following form
(refs. 16 and 17):
03

p(t) = P,~exp(-imBRt) (4)

m=-co
or

where 2PmB is the Fourier transform of the pressure at the mth harmonic of blade-
passage frequency for a propeller with B blades. The term P mis~written as the
sum of effects due to thickness (or volume) displacement Pv,, drag PD,, and lift
P Lso~that
PmB = pVrn + PDm + PLm (6)
Before a noise calculation can be made, the blade geometry and loading must be
specified. A blade planform is specified with the parameters and nomenclature in
figure 11in terms of the chord and sweep as functions of radius ratio z = r/rt, where
rt is the tip radius. Chord b is given by BD,the ratio of chord to diameter D,and
sweep by MCAID, the ratio of mid-chord alignment to diameter. Airfoil section
thickness distribution is specified by the thickness-to-chord ratio at each radius t b ,
and a typical thickness distribution function H ( z ) is shown in figure 12.

15
Magliozzi, Hanson, and Amiet

Figure 11. Blade planform defining chord b and mid-chord alignment (sweep).

-- 1.0

Figure 12. Shape functions for chordwise distributions of thickness and loading.

16
 Propeller and Propfan Noise

Thickness distribution function H ( z ) is defined to be unity at the maximum

thickness point. Similarly, the lift and drag distributions are given in terms of
the section lift and drag coefficients CL and CD and the chordwise lift and drag
distribution functions fL(z) and fo(z) normalized for unit area, as shown in
figure 12. The terms CL and CD are defined such that the forces per unit spanwise
distance are recovered when multiplied by (1/2)p0&t4?, where po and co are the
ambient density and speed of sound and MT,defined by

is the section relative Mach number, with M, and Mt defined as the flight and tip
rotational Mach numbers, respectively.
With these definitions, the noise harmonics are given by (ref. 16)

where J m ~ )( is a Bessel function, 9 v ,!Po,and 9~ are source transforms, and k,

and k, are wave numbers given by

and q5s is a phase lag due to sweep:

Displacement normal to the planform (face alignment) also produces a phase shift
(ref. 16), but that is usually weak and is not included here. The propeller (or
aircraft) position is given in terms of its altitude or sideline distance y, and the
retarded radiation angle 8 is as sketched in the insert of figure 13 (from ref. 19). In
evaluating equation (8) it may be convenient to use the fact that p o z = yp,, where
p , is the ambient pressure and the specific-heat ratio y = 1.4 for air.
The retarded radiation angle 6 and current (or visual) angle 81 are related by
Magliozzz, Hanson, and Amiet

180
- Visual

100 -
Retarded angle, e. deg

I I I I I I I
0 20 40 60 80 100 120 140 160 180
Visual angle, 81,deg

Figure 13. Relationship between retarded and visual angles. (From ref. 19.)

as plotted in figure 13. Finally, the chordwise thickness and loading distributions are
given in terms of their Fourier transforms:

The above equations apply strictly only in the far field. However, they are
reasonably accurate to within about a diameter of the propeller, particularly for the
upper harmonics. Near-field versions of these equations are available (ref. 18). In the
paragraphs below, some general properties of the radiation equations are discussed,
the influence of blade geometry is explained, and some suggestions for programming
are given.
In equation (8), the integrand can be considered to be the product of source terms
(in the braces) times a radiation efficiency factor J m ~ Bessel
. functions of argument
x and order mB # 0 have the behavior shown in figure 14. They peak for arguments
about equal to order, diminish toward zero for smaller arguments, and oscillate
for large arguments. In equation (8), the argument rnBzMtsin8/(1 - Mxcos8)
evaluated for radiation in the plane of rotation is m B z M t . From this, it can be seen
that radiation efficiency at 6 = 90" is governed by z M t , the blade section rotational
Mach number. The factor sine causes the noise to diminish rapidly toward the
front and rear axes of the propeller and the Doppler factor, 1- cos 8, shifts the
directivity pattern forward.
In the source terms, it is easy to see that the thickness, drag, and lift noise com-
ponents are proportional to the thickness ratio, drag coefficient, and lift coefficient,
respectively. The P terms represent the effect of chordwise noncompactness, that is,
interference at the observer location of signals emitted from various source locations

18
 Propeller and Propfan Noise

Figure 14- General behavior of Bessel function with order not equal to zero.

along the chord. The argument k, can be considered the noncompactnessparameter.

Figures 15 and 16 show examples of 9 for thickness and lift sources.
For low frequency (kz -+0), the effect of chordwise interference vanishes. As kz
increases because of increasing chord, harmonic order, or Doppler effects, significant
interference occurs. For chordwise distributions, figure 15 shows that there is
not much potential for reduction of thickness noise by modifying airfoil thickness
distribution. However, the curves in figure 16 show that the quietest chordwise
loading distributions are uniform and that peaky distributions increase the higher
harmonics.
The exponential factor exp(i4,) in equations (8) and (11) indicates that sweeping
a blade section back along the advance helix causes the noise contribution from that
radius to lag by &. Sweep can cause significant noise reduction via the mechanism
sketched in figure 17 for noise radiated at angle 8. Since only one harmonic at a
time is considered, the noise from each blade element is described completely by its
amplitude A j and its phase q5j in the complex notation A j exp[i(c$j - mBflt)].The
total noise is the sum of contributions from all blade elements,

where the common factor exp(-imBflt) has been cancelled from both sides. This
complex addition is easily visualized as the head-to-tail vector addition, also shown
in figure 17. Because the blade sections are swept back, the phase angles of their
signals lag and the vector addition plot tends to close on itself, representing the
phase interference effect. The amount of noise reduction obtained with blade sweep
depends on observer location and propeller operating condition. However, analytical
studies (ref. 17) and test data show that peak sideline noise can be reduced by about
8 dB with blade sweep for the propfan in figure 2 at a cruise Mach number of 0.8.

19
Magliozzi, Hanson, and Amiet

NACA-OOXX
16

General
aviation props
-20[ %? T m B
20lOglo \*'VI Large
conventional props Biconvex
-30

-40 - Propfan mB = 8 I
Series 16
P
0 \ I
Takeoff Cruise \ I
-50 1 I \ I I 1
0 2 4 6 8 10 12 14 16 18 20
Chordwise wave number, IC,

Figure 15. Reduction of thickness noise from blade element due to chordwise
noncompactness. (Prom ref. 17.)

I n s t r o n g peak

-30 - Propfan mB = 8
Takeoff
-40 I I I I I
0 2 4 6 8 10 12 14 16 18 20
Chordwise wave number, k,

Figure 16. Reduction of loading noise from blade element due to chordwise
noncompactness. (From ref. 17.)

20
 Propeller and Propfan Noise

I
/'
..
Vector addition
,/-----

\
\
I \
1

Resultant

Figure 17. Conceptual benefit of blade sweep for reducing noise. (From ref. 17.)

Programming equation (8) is straightforward. However, a few comments are in

order. Since the chordwise distribution needed to compute P is not too critical
for thickness and may not be well-known for loading, it is convenient to use simple
analytical functions for H ( z ) , f ~ ( x )and
, fD(z) that can be transformed analytically
in equation 13. For example, a parabolic thickness distribution,

transforms to

and a uniform lift distribution

21
Magliozzz, Hanson, and Amiet

transforms to
1 (kx = 0)
%=( sin (+) (kx # 0)
These transforms should be accurate enough for most work at low harmonic order.
The need for more accurate source functions can be judged from figures 15 and 16.
In the integration of equation (8), the radial step size must be chosen with some
care, particularly for swept blades, since phase variations along the span increase
with sweep and harmonic order. A simple expedient is to try a step size and see if
the integral is adequately resolved.
Note also that the lift and drag forces defined here are the section loading
components acting parallel and perpendicular to the local section advance direction
as sketched to the left in figure 18. If a lifting-line theory is used to obtain the
aerodynamic loading as input to the noise equations, the reference direction for CL
and CD will probably be shifted from the advance direction by the induced angle,
as shown to the right in figure 18. Correction for this shift should be made before
equation (8) is used.

MZ

The arguments of *L and J,B are unchanged.

Furthermore, if there is no sweep and if *L is taken as real, as in the case
of symmetrical loading functions fL and fD, then an effective radius version of
equation (19) is obtained by setting dT = T and dQ = Q :

22
 Propeller and Propfan Noise

where we have dropped the i and multiplied by 2 so that ion (20) represents
the ordinary Fourier coefficient (one-sided), as denoted by me on P. In terms
of shaft power W = Qfl, the term in the parentheses which contains Q becomes
W / ( z &M;c,), where z,ff is the effective radius, which experience shows can be taken
as 0.8.
To calculate SPL, equation (20) is divided by .\/s to obtain rms pressure and by
the acoustic pressure reference, 4.1784 x lb/ft2 (20 pPa):

538 673mBMt sin 8

SPL = 20 log10
y D ( l - M,cos8) T - z&Mfco
550 w)'YrJmB] (21)
where it is now understood that the thrust is T pounds, the power is W horsepower,
the sideline distance is y feet, the diameter is D feet, and the speed of sound is
e, ft/sec. The Bessel function is

and the noncompactness factor for rectangular loading is

sin X
*L = -
X
where

and 8 is the retarded radiation angle.

Unsteady sources
When the flow into a propeller is nonuniform, blade loading is unsteady and a
significant increase in noise is likely to occur. Because of radiation efficiency effects
described below, a small amount of unsteady loading can be the dominant noise
source, particularly for low-tip-speed propellers. These effects were first recognized
by Lowson (ref. 20), Wright (ref. 21), and Morse and Ingard (ref. S ) , and these
authors give extensive analyses, particularly for helicopter rotors and single rotating
propellers at zero forward flight speed.
Noise caused by unsteady loading can be computed with either time-domain
or frequency-domain methods, and both have their merits. Farassat's formula
(eq. ( 3 ) ) applies without modification for unsteady loading. To use it, the loading
history or waveform 1, and its distribution over the blade must be determined or
approximated. The proper values for 1, are then applied at the appropriate retarded
blade locations. This procedure is reasonably straightforward and Farassat's program
accepts unsteady-loading input. However, there are some subtleties regarding
required quality of the unsteady-loading input data that must be mentioned. Because
the motion of the blades shifts the source frequencies, the blade-loading waveform
must be specified with adequate precision to frequencies significantly higher than
the sound frequency of interest. Since the blade unsteady loading is seldom known

23
Magliozzi, Hanson, and Amiet

accurately, guidance is needed for choosing the time and space resolution of the
surface pressure input. The frequency-domain formulas discussed below may be
used to provide such needed guidance.
Frequency-domain formulas for unsteady-loading noise appear to be more cum-
bersome than their steady-loading counterparts. However, they may be more con-
venient than time-domain formulas for users who are basically interested in noise
harmonics rather than waveforms or who can approximate blade-loading harmon-
ics without the direct time information. Furthermore, the harmonic formulas can
provide valuable diagnostic information because of the frequency discrimination of
spectrum analyzers. Hanson has derived radiation formulas (ref. 22) for the general
case of harmonic blade loading at any frequency, whether or not it is related to the
propeller rotation speed. Two special cases are presented below.
For unsteady loading, the lift coefficient can be expanded in harmonic form as
follows:
00

CL = CLkexp(-iwkt) (25)
k=-w
A similar expression can be derived for CD. Equation ( 2 5 ) gives the lift history
experienced by a blade on the radiating rotor in terms of the lift harmonics CLk,
where k = 1 corresponds to the fundamental frequency and k = 0 is the steady, or
mean, loading designated simply CL previously. For the general case of radiation
from a rotor with angular speed 0 2 interfering with the flow field of another rotor
with B1 blades and rotating at 0 1 , the load frequencies on the radiating rotor are
Wk = kBl(R1 + 0 2 ) . One special case is for counterrotating propellers with equal
speeds (01 = 0 2 = 0) and equal numbers of blades (B1 = B2 = B ) . For this
condition Wk = 2kB0, where the factor 2 arises because of the relative speed of the
rotors. The other special case is for interference with a nonrotating distortion field,
where effectively 01 = 0 and B1 = 1 so that Wk = k 0 .
The far-field acoustic pressure (ref. 22) is

where, for counterrotation with B1 = B2 = B and 01 = 0 2 = 0, the mode order is

n = (m - 2k)B (27)
the wave numbers are

24
 Propeller and Propfan Noise

and the phase lag due to blade sweep is

2BMt m MCA
4 . = - [ My l-M,cose -2k]

For interference with fixed distortion, the mode order is

n=mB-k (31)
the wave numbers are

and the phase lag due to sweep is

In equation (26), the first exponential gives information on the general nature
of the sound field. The frequencies that appear are given by mBR, which indicates
sound at blade-passage frequency and its multiples, just as with the steady-loading
formulas discussed previously. The factor ( T I C o ) - t indicates waves propagating
radially from the source location at the ambient speed of sound e,. The double
summation shows that each loading harmonic k radiates at all the sound harmonics
mB. Variation of the sound pattern in the circumferential direction is given by n4,
where n is the circumferential mode order. These modes, each with n lobes, spin
about the propeller axis with angular speed mB/n times the propeller shaft speed
R. This spinning mode characteristic also occurs for compressors and turbofans, as
discussed in the chapter on turbomachinery noise.
As with thickness and steady-loading noise, the radiation efficiency is governed
by the ratio of argument to order in the Bessel function:

The term E is similar to the cutoff ratio in turbofans. For 5 < 1, radiation is
inefficient; for ( > 1, radiation tends to be efficient because of the Bessel function
properties shown in figure 14. Since zsinO/(l - M, cos e) is of order 1, it is useful
to consider the quantity
Ms = (mB/n)Mt (36)
This is the spinning-mode tip Mach number. Any mode with Ms << 1 can be
neglected in the calculation. Thus, although the summation on k in equation (26)
runs formally from -m to 00, only a few modes for the largest lMsl need be
considered. At the lower sound harmonics sometimes only one mode is required.

25
Magliozzi, Hunson, and Amiet

As an example of the above mode criterion, consider the m = 3 harmonic of a

counterrotating propeller at Mt = 0.7:

mB m
M - Mt =
m - 2k Mt (37)
~

- (m - 2k)B

The combinations of interest for m = 3 are

k m/(m- 2k) lMsl

-1 0.6 0.42
0 1.o .70
1 3.0 2.10
2 -3.0 2.10
3 -1.0 .70
4 -.6 .42
Thus, only the k = 1 and 2 loading harmonics radiate efficiently at 3BPF. ' The
values k = 1 and 2 correspond to blade loading at 2BPF and 4BPF. Through use
of the above formulas, it is easy to show that for increasing sound harmonics, more
and more loading harmonics radiate efficiently. Noise from low-tip-speed propellers
with any source of nonuniform inflow is inevitably dominated by unsteady-loading
sources at the upper harmonics because of their greater radiation efficiency.
Some insight can now be gained for the direct use of time-domain formulas
for calculation of noise from unsteady loading. In equation (3), the unsteady
blade loading is input numerically in discrete time steps. The size of these steps
must be small enough to ensure a full and smooth representation of the loading
component; otherwise, the loading signal will appear to have a higher frequency
that will be strongly emphasized because of the radiation efficiency discussed above.
This sensitivity is aggravated by the derivatives denoted by the dots on li and Mi
in equation (3). The derivations must be performed numerically, a process always
sensitive to smoothness of the quantity being differentiated. These problems are all
manageable in principle. However, the reader is cautioned against casual application
of equation (3) to the unsteady-loading problem without a thorough understanding
of the numerical subtleties.

Nonlinear Effects
Blade sections of propfans and of many other high-speed propellers operate at
transonic velocities. In the aerodynamics of wings and bodies, this is a regime
frequently dominated by nonlinear effects. The corresponding propeller issues are
discussed below under the categories of nonlinear source effects from the acoustic
analogy quadrupole and full aerodynamic solutions by applying finite element
methods.
At high speeds, nonlinearity may occur at the source (i.e., at the blade section)
because of transonic effects. One way of dealing with this is via the quadrupole source
term in the acoustic analogy. (See chapters on jet noise for definition.) The first
valid analysis of the importance of the quadrupole source for high-speed rotors was
made based on a two-dimensional aerodynamic calculation (ref. 23). The quadrupole

26
 Propeller and Propfan Noise

contribution was compared with the linear thickness sourc r a propfan airfoil
section; the results are shown in figure 19. These results show that quadrupole, or
nonlinear, source effects are important sources of additional noise in the blade section
speed range between critical Mach number (when flow over the airfoil exceeds the
speed of sound) and a Mach number of 1. Quadrupole strength can be reduced
to below that of linear thickness and loading sources by blade sweep so the airfoil
sections operate effectively below their critical Mach numbers. To shed more light
on the role of the quadrupole term in flow with solid surfaces, Blackburn examined
the field of a two-dimensional wedge in fully supersonic flow (ref. 24). He was able to
compute the acoustic analogy source terms exactly and found that the quadrupole
was not a significant source of extra noise in this flow regime. This finding agrees
with figure 19, since the Mach’number in Blackburn’s analysis is well to the right
of the peak. However, Blackburn did find that the quadrupole term repositioned
wave fronts along the shocks, rather than on the Mach surfaces as in linear theory.
Quadrupole sources have been treated more extensively in the helicopter literature
(see the chapter on helicopter rotor noise) because supercritical blade section speeds
are common. However, for propellers and propfans, nonlinear effects are minimized
by blade designs with sweep and thin airfoil sections.

Noise added by
quadrupole, dB

Section relative Mach number, M,

Figure 19. Increase of blade thickness sound pressure level caused by

including quadrupole noise. (Prom ref. 23.)

Another approach to nonlinear effects is as a by-product of an aerodynamics

calculation. There is considerable work on the transonic regime in progress to
develop numerical methods for aerodynamic design and analysis from full-potential,
Euler, and Navier-Stokes equations. For the steady-loading problem, it is tempting

27
Magliozzi, Hanson, and Amiet

to believe that since these methods compute all the flow-field variables in a finite
domain surrounding the propeller, the noise signal is available simply by sampling
the pressure field at the appropriate locations. This, in principle, is true. However,
the currently existing aerodynamic analysis methods are designed to give the best
accuracy on the surfaces of the blades. For field points at a distance from the blades,
predictions are degraded because the mesh sizes used in the calculation increase,
numerical damping smooths the waves, and the boundary conditions at the outer
edges of the computational domain are treated approximately. These problems all
appear to be manageable for the steady-loading aerodynamics problem, but so far
they have not been addressed for acoustic calculations. However, recent advances in
computational fluid dynamics are now making this approach look more practical.

Prediction Met hods for Propeller

Broadband Noise
The fundamental laws of acoustics (see Curle, ref. 25, for example ) state that the
noise from a surface is produced by forces (dipoles) and sources (monopoles) induced
on the surface to satisfy the condition of no flow through the surface. If these terms
are harmonic, the noise produced is harmonic. However, if dipoles with a random
time behavior are present on the surface, broadband noise is produced. (Usually
there are no monopoles with a random time behavior, since this would require the
surface to have a significant component of random fluctuation in position.) Random
forces can be induced by several mechanisms. If significant turbulence is present in
the mean stream, random forces are induced on the blades, leading to broadband
noise. In the low-frequency (compact) case the entire blade is involved in the sound
generation process. At higher frequencies (acoustic wavelength smaller than the
chord), the noise generation becomes concentrated around the leading edge of the
blade.
In the absence of inflow turbulence, any random surface forces must be self-
induced. A turbulent flow moving over a plate induces unsteady surface pressures.
For a uniform mean flow this turbulence can be produced in the turbulent boundary
layer. If the turbulence is not in the vicinity of an edge it produces quadrupole
sound, which is generally a weak generation mechanism compared with dipole sound.
However, as the turbulence approaches and passes the trailing edge, the boundary
conditions imposed on an eddy change: whereas the airfoil surface can support a
force, the wake cannot. The result is a change in the airfoil loading as each eddy
passes the edge, and sound is produced.
Two approaches to the calculation of trailing-edge noise have been developed.
Since Curle has shown that the pressure field produced by the turbulence can be
represented by volume quadrupole sources together with the surface monopoles and
dipoles to satisfy the boundary condition on the surface, the first approach is to solve
the problem of a quadrupole in the vicinity of a half-plane. Since the surface dipoles
induced by the quadrupoles are the main sound-producing sources, this method can
be described as a calculation of the surface forces produced by the quadrupoles,
followed by the calculation of the noise. An analysis following this approach was
carried out by Ffowcs Williams and Hall (ref. 26). This approach is satisfactory if
the quadrupole strengths are known. However, this method presents the same kind of
problems encountered in the prediction of jet noise from quadrupole distributions; in

28
 Propeller and Propfan Noise

general the distribution is not known with sufficient accura thermore, there is
the possibility that other nonlinear effects, such as wave ste (also represented
by quadrupoles), axe included with the sound-producing quadrupoles.
These problems led to the formulation of a second approach, which has been more
extensively compared with experiment. In contrast to the above method, this new
method assumes that the surface pressure produced by the convective turbulence is
known. This surface pressure field is assumed to be frozen and convecting with a
velocity U, (or Mach number M,) which may be a function of frequency. At the
trailing edge, there is a change in the boundary condition; in particular, the surface
pressure due to the turbulence, which is supported by the airfoil upstream of the
trailing edge, is not supported by the downstream wake. This produces a fluctuating
dipole force on the surface and radiates sound. Chase (ref. 27) was one of the first to
employ this method for noise prediction. He assumed zero Mach number, and thus
no Kutta condition is applied at the edge. A more general formulation, including
a mean-flow Mach number M and the application of the Kutta condition at the
trailing edge, was introduced by Amiet (ref. 28). This model is especially attractive
because of its symmetrical relation with leading-edge noise. For leading-edge noise,
one specifies the velocity of the incident turbulence field together with a no-flow
condition through the airfoil surface; this leads to zero potential on the axis ahead of
the airfoil and at the leading edge. For trailing-edge noise one specifies the incident
surface pressure on upper and lower surfaces; the pressure difference is zero on the
axis downstream of the trailing edge and at the trailing edge because of the Kutta
condition. However, just as for the first method, this approach does not result in
a compact dipole type directivity pattern. Because the fluctuating forces occur at
the airfoil edge, one cannot simply replace them by the field of a compact dipole in
an infinite fluid. Rather, one must include the baffling effect of the airfoil edge, the
result being a modified directivity.
This model avoids the necessity of specifying the volume distribution of
quadrupoles, since their effect is already included in the convecting surface pres-
sure. Although the specification of the surface pressure is probably simpler than
the specification of the volume quadrupoles, it is by no means a simple task. Amiet
used surface pressure data for a flat-plate boundary layer in his calculations (ref. 28)
partly because this is a simple and “classic” case. Perhaps more importantly, it was
the only readily available data. (When using the following expressions for the noise,
the reader should be aware that the expressions used for the surface pressure are
generally obtained by curve fitting of data. Thus, they are by no-means rigorous and
are open to improvement.)
The far-field noise spectrum prediction Spp above a flat plate for this model is
(ref. 28)

where x,y, and z are, respectively, the coordinates in the axial direction, normal to
the airfoil, and along the span; b is the semichord; s is the semispan; co is the sound
speed; 1, is the spanwise correlation length, which can be a function of the radian
frequency w ; c2 E x2 + ,B2z2,where ,B2 = 1 - M 2 ; C is the generalized lift; and S,
is the pressure spectrum on either the upper or lower surface of the airfoil near the
trailing edge. (To find the total noise, the sound from both surfaces must be added.

29
Magliozzi, Hanson, and Amiet

Since the surface pressure spectra on the two surfaces are generally uncorrelated, the
pressure spectra, rather than the pressure itself, are added.) In the high-frequency
limit (wavelength much less than chord), the directivity factor is given by

with
2 cos2(ee/2)
DE (40)
[I - ( M - M C ) c o ~ ~ , ] 2-( McosO,)
1
where K z = wb/Uc and re and 6, are, respectively, the retarded radius from the
observer to the source point and angle of the observer measured from the upstream
axis. The terms re and Be are related to the actual coordinates by

A more general form of equations (39) and (40) for arbitrary frequency can
be found (refs. 28 and 29), but since trailing-edge noise is predominantly at high
frequency, the more general expression may not be needed. The spanwise correlation
length 1, is defined in terms of the surface pressure spectrum as
1 Po3

The integration of a result from Corcos (ref. 30) leads to an expression for I, of
(ref. 28)
(43)
For a flat plate an expression for S,, can be obtained by curve fitting the data of
Willmarth and Roos (ref. 31) to give

s' =. s9,
-
-
0.00002
(44)
" i(p,U2)2(S*/U) 1 +Z + +
0.217Cj2 0.00562Z4

where po is the free-stream density, Z = wS*/U (a form of the Strouhal number), and
S* is the displacement thickness of the turbulent boundary layer, given approximately
by (ref. 32)
S*
-x 0.047Rc
-115
C
(45)
where Rc is the Reynolds number based on chord e. A reasonable value to use for
the convection velocity is Uc = 0.8U (ref. 28).
Equations (38) to (45) allow a calculation of the trailing-edge noise as a function
of frequency and observer position, given values for U , p,, c,, chord, and span.
One should be able to model trailing-edge noise for a realistic airfoil if accurate
expressions for the surface pressure are known. Of the above relations, the one
most subject to question is equation (44) for the spectrum of the convecting surface

30
 Propeller and Propfan Noise

pressure. Although this equation may be adequate for a flat plate at zero incidence,
one realizes at once that this cannot be quite correct for an actual airfoil, if only
because it does not include the angle of attack. As one might expect, experimental
evidence shows a noise increase as an airfoil becomes more heavily loaded. This
could be due to an increase in either S,, or 6*. Also, the measurements from which
equation (44) was obtained were not made near an edge. For these reasons, attempts
have been made to modify equation (44) with actual airfoil surface pressure data.
Thus, curve fitting the data of reference 33.gives the following general form first used
in reference 34:
I 0.000666G
sqq 1 - 5.4896 + +
36.74G2 0.1505G5
In the range of interest this gives a value for S;, that is somewhat larger than that
given by equation (44) for a flat plate, the maximum difference being 7.7 dB at
G = 0.18. The curve fit is based on measurements taken on an airfoil at zero angle
of attack. However, the surface pressure data are not extensive enough to assure
an accurate prediction. For example, by curve fitting the data of both references 33
and 35, Chou and George (ref. 36) subsequently gave an expression for S;, for two
ranges of G. The first expression, for 3 < 0.06, is taken from equation (46) but is
increased by an overall factor of 2.6. This difference points out the uncertainty in
the result. Although Chou and George used a different curve fit than equation (46)
for the surface pressure, the basic model used for the noise calculation is that of
references 27 and 28, in which a frozen surface pressure convecting past the trailing
edge is assumed.
Obviously the problem is not solved completely until an analytical solution for the
surface pressure is available. However, the above method does give an approximation
to the noise produced if the surface pressure is known; it also gives an insight into the
mechanism of noise production, since it relates the unsteady surface pressure to the
noise produced. There is another prediction method available that dispenses with
the theoretical development and concentrates on curve fitting of available trailing-
edge noise data (in contrast to the above method, which curve fits surface pressure
data and uses this for predicting the noise). This other prediction method uses the
frequency dependence of reference 37 together with certain of the above results and
gives a prediction of 1/3-octave band frequency. The result, with further details
given in reference 34, is

SPL1/3 = OASPL + 10 loglo { 0.613 (

Wmax
7 ) 4 [ ( ~W)m3a’x 2 + o . 5 ] - 4 } (47)

where the overall SPL is

OASPL = 10 loglo M 5 -D
( :; ) + Kl

and s is the span, Gmax is the value of 3 at the spectrum peak (usually around
O.l), and K1 = 141.3. Based on comparisons of theory versus experiment (ref. 34),
equations (47) and (48) give a slightly better prediction than equations (38) to (46),
but equations (47) and (48) give little insight into the noise-generating process.

31
Magliozzi, Hanson, and Amiet

Equations (47) and (48) should give reasonable predictions for an airfoil with a
small angle of attack. F'urther analysis and additional experiments are required to
determine the effects of angle of attack, flow separation, etc., on the noise generation.
(A more detailed analysis of the fundamental principles on which the convecting
surface pressure model is based is given in ref. 28.)
For a propeller, an integral must be taken over the rotor span (i.e., the blade is
treated in a stripwise manner), since each radial station moves at a different velocity.
Because trailing-edge noise is random and generally has a higher frequency than the
propeller rotational frequency, the overall noise spectrum for a rotating blade can
be calculated by simply averaging around the azimuth the sound spectrum derived
above for the case of rectilinear motion. The principle is quite straightforward,
but the implementation can become somewhat involved because of the constantly
changing observer position in blade-fixed coordinates (ref. 34).

Propagation Effects

The propeller noise theories previously described provide estimates of noise

generated at the source without regard to any propagation effects other than spherical
spreading. This section discusses the effects of Doppler frequency shift, refraction,
scattering and shielding, atmospheric absorption, ground reflection, and excess
ground attenuation.

Doppler Frequency Shift

This phenomenon results in a shift in the frequency perceived by an observer

when the observer or the source is moving relative to the medium. The familiar
train whistle is often used as an example. To an observer standing near the train
tracks, the train whistle appears raised in pitch as the train approaches, and lowered
as the train passes by. Similar effects can be observed in propeller noise, the most
important of which is related to the measurement of airplane flyover noise.
For a moving source, stationary observer, and stationary medium the Doppler
frequency shift is calculated from

where fo is the observed frequency, fs is the source frequency, Mx is the flight Mach
number, and 8 is the angle between the line from the source to the stationary observer
and the flight path at the time the sound was emitted. This equation clearly shows
that for an approaching source (8 < 90') the observed frequency is raised, while for
a receding source (8 > 90') the frequency is lowered.
A generalized derivation for the Doppler-shifted received frequency is given by
reference 38. This derivation shows that when the medium is in motion but both
the source and the receiver are stationary (as in a wind tunnel), no frequency shift
occurs. For airplane flyover noise, significant differences in received frequency are
expected for a case with wind compared with that for the zero-wind case.

32
 Propeller and Propfan Noise

Refraction Effects

Propagation Through the Atmosphere

Refraction generally occurs for sound propagating through the normal atmo-
sphere because the temperature and pressure vary with altitude and present a varying
characteristic impedance pc. Although the process is continuous, it is convenient to
approximate it as layers having discrete interfaces (fig. 20). For a discrete interface,
rays are bent according to Snell's law, described in reference 39 as

where the incident and transmitted angles 61 and 62 are defined in figure 20 and
c1 and c2 are respectively the speed of sound in the incident and the transmitted
medium.

pl = A1 exp[i(wt - k l d ) ]

Source

t
+=c Layers

Figure 20. Refraction of acoustic wave.

Some of the energy is reflected, while the rest is transmitted. The ratio of the
amplitude of the transmitted wave to that of the incident wave is

For an acoustic plane wave, the change in sound pressure level is 10 log (Intensity
ratio), where the intensity is A2/pc. It follows that the change in the transmitted
sound pressure level is

33
Magliozzi, Hanson, and Amiet

Because the sound beam is either widened or narrowed when transmitted through
the interface, equation (52) does not give the ratio of total power transmitted. The
change in sound power level can be calculated from

Two special cases bear mentioning. For p2c2 cos 81 = p l c l cos 02, no power is
reflected and all the power is transmitted. When cl < q ,there is a critical incidence
angle Oc, given by sin& = cl/c2, for which the refracted ray is parallel to the
interface. For incidence angles equal to or greater than BC no acoustic energy is
transmitted into the second medium. For sound propagating from high altitudes
through a normal atmosphere, the critical angle can be exceeded at large angles and
no sound would be detected at the ground. This can be significant for propellers
in the forward directivities because the Doppler effect shifts the source directivity
forward.
It is convenient to use discrete layers, typically 100 to 300 m thick, for calculating
refraction effects. Each layer is assumed to have uniform impedance represented by
the mean impedance of the layer. This procedure is recommended for correcting
airplane flyover noise during certification (ref. 40). It is essential to use a layered
atmosphere model for propagation to the ground from high-altitude (greater than
5000 m) flight, as the cumulative effects become significant.

Propagation Through a Fuselage Boundary Layer

Refraction also occurs when sound propagates through a fuselage boundary layer
because the velocity and temperature gradients in the boundary layer cause a change
in the impedance encountered by the sound wave propagating through it. This effect
could be important for noise impingement on a fuselage when cabin noise is being
investigated or controlled.
Several analyses exist for evaluating fuselage-boundary-layer refraction effects
as applied to propeller and propfan noise. Early investigations (refs. 19 and 41)
addressed plane waves and two-dimensional boundary layers. Later refinements
extended the analyses to propeller-type noise sources and boundary layers on
cylindrical surfaces (ref. 42).

Scattering and Shielding

Fuselage Scattering
Sound incident on a cylindrical fuselage is scattered depending on the angle
of incidence and the wavelength of the sound compared with the diameter of the
fuselage. For normal incidence, sound at a small wavelength compared with the
fuselage diameter is totally reflected. Thus, a receiver at the fuselage surface
perceives a doubling of the pressure.
The analysis of reference 42 represents the fuselage as an infinitely long cylinder
with infinite impedance at the surface. Scattering effects from this analysis are shown
in figure 21 for the fundamental and second harmonic of a model propeller. On the

34
 Propeller and Propfan Noise

side with the incident sound, pressure doubling is shown at both frequencies because
the wavelengths of both tones for the model propeller are sm
fuselage diameter. On the opposite side, a shadow zone is
shadow occurring for shorter wavelengths. Interference is shown at the edges. Note
that the pattern appears rotated (i.e., the pattern is not symmetric about q3 = 0')
because of the rotating source. Thus the analysis is sensitive to the direction of
rotation of the propeller.

0.5 diameter aft Plane of rotation . 0.5 diameter forward

Figure 21. Calculated fuselage scattering effects at MZ = 0.8. (From ref. 42.)

Wing Shielding
As is the case for a fuselage, a wing can be used to provide shielding of a propfan
source. Figure 22 illustrates the situation for a propfan installed on a swept wing.
It is shown that for the geometry and the direction of rotation indicated, the line
of sight from the advancing blade is blocked by the wing leading edge. As is the
case for the fuselage, the amount of shielding depends also on the wavelength of the
sound, with shorter wavelengths approaching geometric acoustic behavior.
Several analyses have been developed for shielding of sound by stationary barriers.
These analyses were extended for propfans installed on wings by including the effects
of flight Mach number and wing sweep in reference 43.

Atmospheric Absorption
When sound propagates over long distances through air, absorption takes place
and reductions in amplitude in excess of those from simple distance effects are
observed. These effects have been studied for some time and several procedures
exist to calculate the effects (refs. 44 to 47). The method endorsed by the FAA
and recommended for adjusting noise certification data is that of reference 47.

35
Magliozzi, Hanson, and Arniet

?uselage

>Shielded region

(a) Top view.

Direction
of

propagation
to fuselage

(b) R o n t view.
Figure 22. Shielding of propeller noise by swept wing.

36
 Propeller and Propfan Noise

This procedure presents an algorithm which allows the calculation of atmospheric

attenuation, in terms of a dB attenuation per unit distance, as a function of
temperature and relative humidity. These values typically vary from 0.001 dB/m
at low frequencies to about 0.1 dB/m at high frequencies. Because the atmospheric
attenuation has relatively little effect on low frequencies, it does not have a strong
influence on propeller noise, except for the case of noise propagating to the ground
from high altitudes (above 5000 m).

Ground Reflection Effects

In a typical airplane noise measurement situation, the airplane flies past a
microphone which is located above a ground plane. The sound thus reaches the
microphone following the direct path and a reflection from the ground plane, as
illustrated in figure 23 for a simple point source S. Also shown is the equivalent
image source SI, which accounts for the ground reflection process. The distance
traveled by the direct ray R is given by [L2 + ( H - h)2]1/2, whereas that of the
reflected ray R‘ is given by [L2+ ( H +
Note that R’ is always longer than
R, except in the special cases of H = 0 or h = 0, for which they are equal. Since the
propagation lengths along the two paths differ, the signals arrive at the microphone
with relative phase differences that cause constructive interference, when the two
signals are in phase, or destructive interference, when the two signals are out of
phase.

Ground
plane
0
0
0
0
0
0

0’ R’ A R = R’-R
0
0 a = 2?r AR f l c
0
0 For R‘ % R and an infinite
0
0 impedance ground plane: ASPL = 10 log(2+2 cos a )
0
0

0’
S‘

Figure 23. Ground reflection effects with image source.

37
Magliozzi, Hanson, and Amiet

For noise certification purposes, a microphone height of 1.2 m is required. At

normal ambient temperatures, this height results in cancellations at approximately
80 Hz for a source directly over the microphone ( L = 0). This value is in the range of
blade-passage frequencies for many propellers. Since propeller noise is predominantly
at low frequency, these effects can thus be significant.
It is difficult to interpret measurements made using a microphone located above a
ground plane because of the changing ground reflection effects coupled with apparent
source characteristics (e.g., directivity and Doppler shift). One method which gives
good results is to use a microphone close to ground level ( h M 0) over a hard surface
(for high impedance). This arrangement results in the reflected signal always being
in phase with the direct signal, so a nearly constant 6-dB correction (full pressure
doubling) occurs over the frequency range of interest, independent of source position.
Ground reflection effects are needed for estimating the noise to be expected during
noise certification, for instance. Procedures for calculating ground reflection effects
can be found in references 48 to 53 and are discussed in another chapter of this
book. These methods do not necessarily address tone sources. Ground reflection
corrections for propeller harmonic noise should be done for small bandwidth signals
at the Doppler-shifted tone frequencies. Using center frequencies of 1/3-octave bands
can lead to significant errors. Experience has indicated that adjusting a ground
microphone measurement of a propeller aircraft flyover to 1.2 m cannot be done
with high accuracy with existing methods because of the complexity of the ground
reflection process.

Excess Ground Attenuation

Excess ground attenuation, sometimes called lateral attenuation, is a term
applied to discrepancies between observed levels and those expected after all other
propagation effects (i.e., distance, atmospheric attenuation, and ground reflection
effects) have been accounted for. This effect is usually found when measurements
of flyovers are compared with those from a sideline microphone. A compendium of
such measurements has been published in reference 54. A summary of these results is
shown in figure 24. It is probable that a significant portion of the lateral attenuation
shown in figure 24 is a result of shielding because of the apparently stronger effects
for fuselage-mounted engines. It is not clear how this effect would be reflected in
propeller noise. However, this can certainly be best addressed by applying wing and
fuselage shielding and scattering analytical methods.

Nonlinear Propagation Effects

Open rotors generally produce intense noise levels. This is particularly true of
propfans during high-speed cruise. Under these circumstances, significant nonlinear-
ities can arise. There may be nonlinear propagation effects in addition to the nonlin-
ear source effects discussed previously. Nonlinear propagation of propeller noise was
first studied by Hawkings and Lowson (ref. 55). Since then, Barger (ref. 56), Tam
and Salikuddin (ref. 57), and Lindblad (ref. 58) have pursued the subject further.
All these investigators applied weak shock theory as developed for sonic booms to
the propeller noise propagation problem. The analysis is applied in conjunction with
a linear source theory calculation, although it is not inherently limited to this and
could be matched with a nonlinear source theory or even with experimental results.

38
 Propeller and Propfan Noise

Observer angle, deg

Figure 24. Typical variation of lateral attenuation with observer angle.

( B o r n ref. 54.)

The approach is to begin with an acoustic pressure waveform and apply a nonlinear
propagation theory to account for the wave steepening that builds up during propa-
gation because of finite signal amplitude. In the positive pressure peaks of the wave,
the temperature and speed of sound are above ambient values so that the peaks
propagate faster than the valleys.
Examples of the nonlinear propagation effects are shown in figure 25 for an
unswept propfan blade. In figure 25(a) the noise pulse was computed from a linear
theory equivalent to equations (3) or (28). Weak shock theory was applied to produce
the wave in figure 25(b). As shown, the nonlinear propagation theory steepens the
leading edges of the pulses and compares better with experimental results (fig. 25(c))
than does linear theory. A shift in energy is made from the lower frequencies to
higher frequencies, although there does not appear to be a significant reduction in
level at the very low frequencies. The effects described herein occur very close to a
propfan. Test data show that propagation of sound from propfans follows the linear
propagation laws of typical acoustic sources at distances greater than one propfan
diameter.

39
Magliozzi, Hanson, and Amiet

k . 1R e v 4 c . 1 ReVJ

(a) Computed from (b) Computed from (c) Test wave-

linear source and linear source and form; Mt =
linear propagation. nonlinear propagation. 1.18.

Figure 25. Nonlinear propagation eflects computed from weak shock theory.
(Born ref. 56.)

Comparisons of Predictions and

Measurements

Background
Before attempting any serious study of the performance of propeller noise
prediction methods, one must fully understand the methods used to make the
measurements. This is important because test facilities may influence the noise-
generating process or propagation in a manner not modeled by the noise prediction
method. Other considerations include the presence of additional sources of noise
in the measurements (e.g., a drive motor) and the assurance that the propeller is
operating at the conditions (i.e., blade loading, relative velocities, etc.) defined for
the calculations.

Noise Measurements Under Static

Conditions
As described previously, a propeller operating under static conditions encounters
a great deal of nonuniform inflow, including naturally occurring turbulence in the

40
 Propeller and Propfan Noise

atmosphere, ground vortices, and wakes from fuselages, wings, nacelles, or test
stands. This inflow results in high levels of unsteady-loading noise that tends to
dominate the higher sound harmonics. This source of noise disappears quickly when
a small amount of forward speed is attained. A passenger in a propeller-driven
airplane may observe high levels of noise up to the point of brake release, but the
noise quickly changes as the airplane reaches a modest speed during the takeoff roll.
Similar effects can be observed on static test stands in that the noise can change
markedly when a modest head wind occurs.
Forward flight effects on propeller noise have been investigated (refs. 59 and 60).
Figure 26 shows representative noise measured during static and flight conditions at
constant propeller speed and power. As shown, the static data are dominated by
high levels in the upper harmonics, but these are essentially gone in the flight data.

SPL,
dB

0 10 20 30
Harmonic of blade-passage frequency

I I I I
0 1000 2000 3000
Frequency, Hz

Figure 26. Effect offorward flight on propeller noise. (From ref. 59.)
Comparisons between measured and calculated levels for static propeller noise
are presented in reference 60. It is concluded that a compact source calculation for
fluctuating blade-loadmg noise and a noncompact source calculation for thickness
and steady-loading noise are adequate for predicting the noise of static propellers.
The prediction of static propeller noise, however, is not of great interest, as the
condition is transient and is not used for noise certification or interior cabin noise
control.

41
Magliozzi, Hanson, and Amiet

The noise from counterrotating propellers operating under static conditions would
probably not show as much influence from turbulence ingestion effects because the
higher harmonics are dominated by aerodynamic interaction between the two blade
rows. Limited data exist (ref, Sl), but it is not clear what components comprise
the noise. It can be concluded that any static propeller data to be used for
correlation purposes should be viewed with caution, as they are strongly influenced
by nonuniform inflow, which is not usually well defined and often varies.

Propeller Noise Measurements in Wind

Tunnels
There are two common types of wind tunnels used in acoustic research: the closed-
test-section type and the open-jet type. Both offer advantages and disadvantages in
regards to propeller testing.

Closed- Test-Section Wind Funnels

Most closed-test-section wind tunnels are of the recirculating type. It is thus
necessary to control the turbulence which can be introduced by the drive fan,
turning vanes, recirculating wakes produced by the model, etc. If the test-section
walls are metal or concrete, many reflection paths can exist, the result of which is
great variability in the measured noise. This variability is particularly significant for
propeller noise because of its discrete frequency components. Depending on mode
characteristics of the tunnel, significant reinforcements or cancellation can occur at
the harmonic frequencies (ref. 62).
One solution to this reflection problem is the use of absorptive treatment on the
tunnel walls. Because of aerodynamic losses caused by flow over the treatment,
the use of wedges is not practical. Flat-faced fiberglass (behind high-open-area
perforated retaining plates) or polyurethane foams have been used. These appear to
work reasonably well at low to moderate speeds (below Mach 0.5) for measurement
locations near the peak noise (ref. 62). This type of treatment might not work well
at high speed or at shallow incidence angles which occur at locations forward of the
propeller plane of rotation. Acoustic qualities of wind tunnels operating at speeds
above Mach 0.5 have not yet been demonstrated.

Open-Jet Wind Tunnels

Another approach is the use of open-jet wind tunnels. In this arrangement, a
nozzle is typically set into the wall of an anechoic chamber. A collector is situated
opposite the nozzle. When suction is applied at the collector a jet forms between the
nozzle and the collector. Placing a propeller in the jet simulates flight. Because there
is essentially no flow outside the jet, the chamber can be treated with acoustic wedges
to provide an anechoic environment. The limitations of this scheme, however, are
restricted speed (about Mach 0.5) and propagation effects through the shear layer
to far-field microphones. Shear layer corrections for amplitude and directivity angles
exist and have been well documented (refs. 63 to 65). The angle corrections show
that the sound is refracted by the shear layer. In the forward direction, the refraction
can be complete so that no sound passes through the shear layer, thus limiting the
range of directivity attainable.

42
 Propeller and Propfan Noise

Propeller Noise Measurements in Flight

Measurements of noise from propellers installed on airplanes in flight should be
the best data for correlation purposes because the ultimate objective of studying
propeller noise is to control it in the passenger cabin and in the communities affected
by noise of airplanes taking off, landing, or in fight. The measurement of propeller
noise from airplanes, however, is difficult. Apart from the generally higher costs
associated with flight tests, propellers installed on airplanes are subjected to inflow
distortion such as angle-of-attack effects, wing upwash, nacelle blockage, engine
inlet effects, and wakes from upstream disturbances which may significantly affect
the propeller noise characteristics. In addition, other sources of noise are present,
such as noise of the airframe 'and of the engines. Finally, the interpretation of
airplane noise should include atmospheric attenuation, ground reflection, shielding,
and Doppler shifts for propagation to the ground and fuselage reflections, refraction,
and propagation through a fuselage boundary layer in the near field. (See previous
discussions in this chapter.)

Comparisons in the Near Field

Predictions made with the Hanson frequency-domain method (ref. 18) and
measurements made on two model propfans (ref. 66) operating over a range of tip
speeds and blade loadings are shown in figure 27. The SR-2 model blades are straight
whereas the SR-3 blades are swept according to recent practice. It is apparent that
the level of the prediction method agrees well with test data, with the trends of noise
versus tip speed being well predicted. Also, the benefits of blade sweep are shown in
both the measurements and the predictions.
Figure 28 shows the measured and predicted directivity of the blade-passage
frequency harmonic. This comparison indicates that all sources of noise, including
the nonlinear quadrupole, are important, particularly at the forward location, when
the relative Mach numbers are high. It is shown that the total noise energy is
generally not equal to the sum of the component energy because of relative phase
effects.
Comparable results are obtained from calculations with one of Farassat's time-
domain methods. Sample correlations (ref. 10) are shown in figures 29 and 30.
As shown, the general characteristics of the waveforms are predicted well by the
time-domain method except for the positive peaks, which are reduced by nonlinear
propagation. This is also shown in the spectrum in figure 29(c). Figure 30(c) shows
better spectrum agreement than does figure 29(c), although the waveform correlation
is not as good. The Hanson frequency-domain method (ref. 18) at BPF shows good
agreement with measured data in both figures. In fact, it is expected that time-
domain and frequency-domain methods would show nearly identical results within
the realm of linear acoustics.

Comparisons in the Far Field

Measurements of the noise from a full-scale general aviation propeller were
made in a large open-jet anechoic tunnel (ref. 67). These data and predicted
values from a time-domain method are shown in figure 31. It is apparent that the

43
Magliozzi, Hanson, and Amiet

Free-field
noise predictions P
1AE L 0 SR-2

BPF
SPL,
dB

105
.7 .8 .9 1 .o 1.1 1.2
Tip helical Mach number

Figure 27: Measured and predicted sideline tone levels versus tip helical
Mach number. Boom microphones; power coeficient, 1.9; advance
ratio, 3.1. (From ref. 66.).

agreement between measurements and predictions for these cases where the propeller
is operating in low turbulence, undisturbed flow is excellent.
Figure 32 shows predictions and measurements for the propeller operating on
an airplane. Measurements were made with a ground-level microphone and the
noise predictions included propeller angle-of-attack effects. As shown, the noise is
underpredicted ahead of the overhead point and overpredicted behind the overhead
point. It is conjectured that this could be caused by nonuniform inflow effects other
than propeller angle of attack.
Reference 68 presents results for a model propeller operated in an open-jet
facility with microphones located inside the jet to avoid shear-layer refraction effects.
In addition to noise measurements, the propeller aerodynamics were measured
to confirm the blade-loading distributions, which are inputs needed to calculate
propeller-loading noise. Representative time-domain measurements and predictions
are shown in figure 33. These plots show very good agreement between measured
and predicted values. Although only waveforms are shown, it would be expected
that there would be excellent agreement of harmonic data as well.
Measured and predicted values for a propfan operating at takeoff and landing
conditions in a large, acoustically treated, closed-test-section wind tunnel are shown
in figure 34 (ref. 69). For these predictions, the aerodynamic performance was

44
 Propeller and Propfan Noise

t
I a Monopole 1
4)Corrected to free field
0 Dipole
SPL, 110 + Quadrupole

dB
100 -

90 -
Plane
80 -
of
rotation

70 I
Aft
I t I
Forward
I I 1

Figure 28. Measured and predicted BPF tone sideline directivity for
SR-3 operating at design point condition. (From ref. 66.)

predicted with an Euler code and the noise was predicted with the time-domain
method of Farassat. Figure 34 shows the predicted and measured noise for two
harmonics as functions of directivity for three blade angles p. Figure 35 shows the
waveforms for the three conditions in the propeller plane of rotation (at or near the
peak directivity angle). There is a tendency to underpredict as the propfan blade
angle is increased. This underprediction is attributed to the formation and increasing
strength of a leading-edge vortex resulting from the thin, sharp-edged, swept blades
at increasing blade angle. The leading-edge vortex, and also an associated tip-edge
vortex, can change the blade-loading distributions significantly. If this change is not
reflected in the blade-loading source distribution, then noise predictions can become
inaccurate.
Noise measurements for a propeller operating on an airplane in flight were made
using microphones mounted on a wingtip (ref. 59). Predictions were made using
a frequency-domain method (ref. 70). Figure 36 shows these values for a propeller
tip rotational Mach number of 0.77, while figure 37 shows these values for a tip

45
Magliozzi, Hanson, and Amiet

(a) Ezperimental.

2000
Acoustic pressure,

(b) Theoretical.
-1000
0
' I ' I "
1
' " " " " "
2 . 3
" '4
Time/Period

0 5 10 15 20 25 30
Harmonic number

Figure 29. Measured and predicted acoustic pressure signatures and spectrum
for boom microphone directly above propeller disk. Theoretical prediction
corrected for boom reflection. Period, 0.991 msec; BPF = 1009.2 Hz.
(&om ref. 10.)

46
 Propeller and Propfan Noise

500
0
-500
-1000

Acoustic pressure, Pa
1000 F

1 ' 1 ~ ~ ' ' ' ' ' '

0 1 2 3 4
150
140
SPL, 130
dB 120
110

loo 0 5 10 15 20 25 30
Harmonic number

Figure 30. Measured and predicted acoustic pressure signatures and

spectrum for boom microphone behind propeller disk. Boom reflec-
tion correction included. Period, 0.991 msec; BPF = 1009.2 Hz.
(From ref. 10.)

rotational Mach number of 0.83. The agreement is quite good in the propeller plane
of rotation and fair at the aft location, although the spectrum shape is well predicted.
No nonuniform flow fields were included in the noise predictions. These comparisons
show the importance of the thickness noise component for this propeller at these
operating conditions.

General Comments
The following general observations are based on the foregoing comparisons
between measurements and predictions of propeller and propfan noise.
Generally good agreement between measurements and predictions of noise can be
obtained for propellers operating at low-to-moderate tip speeds at moderate loadings
under ideal (undistorted) inflow conditions. For these conditions, the prediction
model needs to include only linear sources, and comparable performance can be
obtained with either time-domain or frequency-domain methods. It is apparent that
for good loading noise calculations, the blade-loading distribution must be accurately
defined.

47
Magliozzi, Hanson, and Amiet

110
o FAA/DFVLR DNW test
100 I ANOPP PAS prediction
90
Directivity angle = 60" o FAA/DFVLR DNW test
80 I ANOPP PAS prediction
70
60
SPL.' 50
dB
40
110
100
90
80
70
0 5 10 15 20
Harmonic number Harmonic number

(a) First case; power co- (b) Second case; power co-
eficient, 0.028; Mt = 0.781. eficient, 0.049; Mt = 0.872.

o FAA/DFVLR DNW test

120 r I ANOPP PAS prediction
110 Directivity angle = 60"
100
90

SPL, 8o
dB 70
120
110
100
90
0 5 10 15 20
Harmonic number

(e) Third case; power co-

eficient, 0.040; Mt = 0.876.

Figure 31. Predictions and data for full-scale general aviation propeller in
large open-jet wind tunnel. (F'rom ref. 67.)

48
 Propeller and Propfan Noise

o FAA data
I ANOPP PAS prediction
goroo Maximum measured dBA

Overhead emission
t=O

90
Past overhead
80 t = 1.5
70
60
50
0 5 10 15 20
Harmonic number

Figure 32. Measured and predicted noise spectra at three time periods
for standard takeoff of Piper Lance airplane. (From ref. 67.)

The prediction of propfan noise is generally not as successful. At low speeds

this is not necessarily due to failure of the same noise prediction models as above
(although the existence of other sources such as a tip-edge loading is a possibility),
but more likely it is due to the failure of the aerodynamic model to predict the
blade-loading distribution.
During high-speed operation, additional (nonlinear) sources and/or nonlinear
propagation near the propeller become apparent.
Finally, propellers and propfans installed on airplanes have other sources of noise,
notably unsteady-loading noise, due to inflow distortion. Again, these effects have
been included in many propeller noise prediction models, but they require a means
of defining the unsteady blade loads. Accurately estimating unsteady blade loads
is not easily managed by current aerodynamic methods and generally the resulting
noise predictions are not as good as those for steady blade loads.

Propeller Noise Control Objectives

It is the job of the acoustician to first understand the propeller noise-generating
mechanisms and then to control them using methodologies derived from theories
to meet constraints demanded by airplane manufacturers. These noise constraints
are based on meeting noise regulations, cabin noise comfort, airplane structural
requirements, etc. In the following discussion the general noise control objectives

49
Magliozzi, Hanson, and Amiet

P, Pa o
-20

~~~~ , , , , , ,
-80
-1.5 -1.0 -.5 0 .5 1.0 1.5
Time, msec

(a) -40" azimuth.

E 1 . 5 -1.0 -.5 0 .5 1.0 1.5

Time, msec

(b) 0" azimuth.

80 -
60 -
40 -

-40 -
-60 -
-80 i 1 1 , 1 I I
-1.5 -1.0 -.5 O .5 1.0 1.5
Time, m e c

(e) 40" azimuth.

Figure 33. Measured and,predicted pressure signatures at three azimuths

for propeller on symmetric body. Propeller speed = 10000 rpm. ( R o m
ref. 68.)

50
 Propeller and Propfan Noise

120

110

100
SPL,
dB 90

80

70 I I I 1

- Data
-e-
.
= 37.8'
11203 0 ~
-+-I Prediction

90
\ 4 3 x BPF
80 I I

43.6"
43.6"f i T '
:0% D i -e-
o n

k"
d/
n\ 2 x BPF

80
Flow
I
I 1
0 -100 -200
Axial distance, crn

50
w
70 90 110 130
Traverse angle, 6, deg

Figure 34. Measured and predicted values for SR-7A model. 1.68 m sideline;
advance ratio, 0.886; Mz = 0.2. (From ref. 69.)

51
Magliozzi, Hanson, and Amiet

-
----- Data
Prediction

Sound 7
pressure,
Pa -201 V
-40

-
40

Sound
pressure,
Pa

-40

Sound
pressure,
Pa

-80
0 .5 1.0
Time/Period

Figure 35. Waveforms in plane of propeller at three blade angles. Advance

ratio, 0.886; Mz = 0.2. (From ref. 69.)

52
 Propeller and Propfan Noise

are grouped into two sections. In the first, the near-field noise requirements, which
generally pertain to the cruise condition, are described. Then the far-field noise
requirements, which generally apply to low-speed operation, are discussed.

Near-Field Noise Control Objectives

In a typical installation, the propeller is located adjacent to a fuselage section.
Depending on propeller design, operating conditions, and proximity of the propeller
to the fuselage, the acoustic excitation can produce intense fluctuating pressure loads
on the surface. These loads can cause motion in the fuselage structure, with possible
fatigue of the structure and noise transmission to the interior. This motion can be
particularly important if the structure has response frequencies which coincide with
the propeller noise frequencies. It is therefore desirable to control the propeller noise
levels and to avoid excitation at structural resonances to reduce or eliminate acoustic
fatigue.
For passenger comfort, the propeller noise reaching the fuselage interior space
needs to be controlled. Today’s airline passengers expect cabin comfort in propeller-
driven airplanes to be comparable to that in turbofan-powered airplanes. This means
limiting cabin noise levels to 80 dBA or less. Further, for enhanced comfort the actual
propeller noise harmonics should be barely discernible. This usually implies that the
propeller harmonic noise contribution is below 80 dBA and the broadband noise from
fuselage boundary layer, environmental control system, etc., has comparable levels.

Far-Field Noise Control Objectives

In general, far-field noise control addresses community noise objectives. The most
important of these is noise certification as set forth by the FAA (ref. 40). Additional
noise requirements may be imposed by certain airports for takeoff and landing.
For an aircraft to receive certification it must satisfy noise constraints during
takeoff and landing. For turbojet and transport category airplanes, these are
currently defined by Stage-3 requirements described in Appendix C of Federal
Aviation Regulations (FAR) Part 36 (ref. 40). Comparable requirements are imposed
by the International Civil Aviation Organization (ICAO) in Annex 16 (ref. 71). In
both cases the noise limits are specified at three locations, as defined in figure 38
(from ref. 72). These documents describe the procedure to be followed in certifying
and specify the limits to be met, which depend on the location and the airplane gross
weight.
For airplanes below 12 500 lb gross weight, the certification procedure and limits
are different. These are described in Appendix F of FAR Part 36 (ref. 40). In general,
this certification requires level flyover over a microphone. Adjustments are allowed
for good takeoff climb airplane performance.
Although many airports have noise restrictions of various types, two airports in
the United States are particularly strict. The first of these is Washington National
Airport, which requires low noise for nighttime operation based on results of the
FAR Part 36 certification testing. The limits, however, are based on the maximum
A-weighted sound pressure level. The maximum level for takeoff is 72 dBA, while
the maximum level on approach is 85 dBA. No requirements are made on sideline
noise. Airplanes not meeting these requirements may not take off or land at the
airport between 10 p.m. and 7 a.m.

53
Magliozzi, Hanson, and Amiet

110

100

90

80
SPL,
dB
70

60

50 - X Loading only
- Loading plus thickness
40 I

(a) In-plane location.

110

100
-x Loading
Loading only
plus thickness

SPL, go
dB

80

70

60
0 10 20 30

Harmonics of blade-passage frequency

(b) Aft location.

Figure 36. Measured and predicted propeller noise levels at tip rotational Mach
number of 0.77. (Prom ref. 70.)

54
 Propeller and Propfan Noise

120

110

100

90
SPL,
dB
80

70

60 X Loading only
- Loading plus thickness
50
0 10 20 30
Harmonics of blade-passage frequency

(a) In-plane location.

SPL,
dB

I
0 10 20 30
Harmonics of blade-passage frequency

(b) Aft location.

Figure 37. Measured and predicted propeller noise levels at tip rotational Mach
number of 0.83. (Born ref. 70.)

55
Magliozzi, Hanson, and Amiet

Measurement

Figure 38. Noise certification measurement locations. (From ref. 72.)

The second airport having special noise rules is John Wayne Airport in Orange
County, which is outside Los Angeles, California. For John Wayne Airport, the
measurement units and locations are significantly different from the FAR Part 36
rules. The measurement locations can be seen in figure 39 (from ref. 72). Microphone
locations are scattered within a several-mile radius of the runway.
The measurement unit at John Wayne Airport is single-event noise exposure level
(SENEL). The airport requirements for varying numbers of allowed flights per day
are given below.

Classification SENEL, dBA

Unrestricted, unlimited flights < 86
AA 86 to 89.5
A 89.5 to 100

Airplanes 'meeting the class A level are allowed fewer flights per day from the airport
than those meeting the class AA level.

Control of Propeller Noise

It is possible to obtain guidance in controlling propeller noise by inspection of
the noise prediction theories. For example, in equation (8 ) relative Mach number is
a multiplier of the noise level. Thus, reducing blade section relative Mach number
should reduce noise. In fact, for most cases that is indeed a way to reduce propeller
noise-lower tip speeds almost always reduce noise. Other approaches include

56
 Propeller and Propfan Noise

Noise Monitor Station Locations

Distance, ft
Mic )hone
he It, ft
From north
end (on extended At 90' t o
C/L) runway extended
Station Location 01L-19R C/L 01L-19R AGL ARL
M-1 Newport Beach, CA 10 800 200 Easterly 20 13
M-2 Newport Beach, CA 17470 1220 Westerly 20 57
M-3 Newport Beach, CA 14 800 3570 Easterly 20 67
M-4 Santa Ana, CA 5 850 300 Westerly 43 26
M-5 Tustin, CA 28 650 175 Easterly 22 109
M-6 Santa Ana, CA 8 660 950 Westerly 25 15.5
M- 7 Santa Ana, CA 8 870 900 Easterly 25 24
M-8 Newport Beach, CA 24 200 3030 Easterly 25 7
M-9 Santa Ana. CA 17 700 5350 Westerly 59 106

Remarks:
1. Length of runway 01L-19R: 5 700 ft.
2. Runway elevation: 53 ft mean sea level.
3. AGL = above ground level.
4. ARL = above runway level.
5. Runway magnetic heading: 194' 21'.

Station Location Map

Figure 39. Noise measurement locations for John Wayne Airport. (From ref. 72.)

57
Magliozzi, Hanson, and Amiet

altering operating conditions, changing propeller design, and, for the special case
of cabin noise with multiple propellers on the airplane, using a synchrophaser.

Operating Conditions
As previously mentioned, certain operating conditions can be modified to reduce
noise. The most significant of these is tip speed. As can be ascertained from the
governing noise equations, all sources of noise have radiation efficiencies determined
by relative velocity. It is not straightforward to determine how much the noise will
change with tip speed. In the case of loading noise, as an example, there are three
significant parts of the noise-generating process which are affected. The first, obvious
by inspection, is that relative Mach number multiplies the equation. Mach numbers
also appear in the argument of the Bessel function. In this case, the effect of Mach
number is not as clear, but reference to figure 14 or to Bessel function tables indicates
that for subsonic tip speed and flight speed, reducing the Mach numbers reduces the
noise. Finally, in order to maintain thrust (that is generally a firm requirement,
as the application for propellers is generally to fly an airplane at some speed and
altitude), the lift and drag coefficients change if tip speed is reduced.
Other ways to reduce propeller noise are to increase diameter or to reduce the
disk loading (i.e., thrust per unit area of the propeller disk). In order to maintain
thrust, operation at a lower disk loading requires a larger diameter.
Reducing tip speed has generally reduced noise for all sources. The best way to
evaluate this effect is to perform calculations while observing the established ground
rules (e.g., maintaining constant thrust). This is needed because the benefit depends
on specific designs and baseline operating conditions. As a reference, it has been
observed that for conventional propellers operating at low to moderate flight speeds,
the overall noise in decibels varies as approximately 40 times the tip Mach number
(ref. 73).
Reducing disk loading affects primarily loading noise. Again, evaluating the
benefits requires specific calculations. As a guide, an empirical propeller noise-
estimating method (ref. 73) indicates that noise varies inversely as diameter squared.

Design Parameters
It is apparent from the foregoing discussion that examination of the propeller
noise equations shows areas where noise reduction benefits can be attained. For
example, propeller noise could be greatly reduced by having zero-thickness blades
to eliminate thickness noise, large diameters with many blades to eliminate loading
noise, and large blade sweeps to eliminate quadrupole noise. Unfortunately, practical
realities must also be considered. These include physical constraints, such as a certain
amount of blade thickness needed for structural integrity, and practical constraints,
such as a limit to the diameter for weight and installation considerations. However,
general guidance can be obtained from the equations governing propeller noise.
Although specific benefits must be evaluated individually and in combination for
specific cases, the following is given for general guidance.

Blade Sweep
Increasing blade sweep is beneficial during high-speed cruise, when blade section
relative Mach numbers are relatively high. Figure 40 shows calculated noise reduction

58
 Propeller and Propfan Noise

Near-field noise
reduction, dB

Blade sweep, deg

Figure 40. Noise reduction due to blade sweep calculated using

frequency-domain method. M z = 0.8; tip speed, 8OOft/sec; 8-bladed
propfan; BPF harmonic.

for blade sweep. It is apparent that a small amount of sweep has a small effect on
noise, but sweep becomes very effective at reducing noise as sweep increases. These
effects were calculated for a propfan during cruise. Benefits during takeoff would be
significantly less (ref. 17).

Blade Thickness
Thickness noise is significant during high-speed cruise, when blade section relative
velocities are high. One means for reducing the contributions from thickness noise is
to reduce the blade thickness. Actually, reductions are obtained by reducing blade
thickness and chord, as it is the blade volume which factors into ‘the source strength.
The effect on spectrum depends on the shape of the airfoil. A scaled reduction in
airfoil thickness at constant chord provides reduction equally at all harmonics. The
noise reduction attainable varies as approximately the blade volume squared.
Reducing blade thickness also reduces quadrupole noise, but in a less predictable
manner.

Blade Count
For a given thrust requirement, increasing blade count is always beneficial in
reducing loading and quadrupole noise. Thus, at low-speed takeoff conditions, where
loading noise dominates, reduction is obtained by increasing blade count. Although
significant reduction in noise level (particularly at the higher harmonics) occurs,

59
Magliozzi, Hanson, and Amiet

some of this reduction may be offset by increases in metrics which are frequency
dependent, such as effective perceived noise level. This increase occurs for a given
tip speed and diameter because increasing blade count raises the frequencies. In
general, though, a net noise reduction can still be obtained.
Increasing blade count can raise thickness noise, depending on how it is done.
Examination of equation (8) shows that simply adding blades increases the number of
sources of thickness noise, with a corresponding increase in noise. If the blade volume
is decreased (by reducing chord), then increasing blade count may not have as much
effect on thickness noise. Again, adding blades raises the frequencies generated,
so that metrics such as A-weighted overall levels commonly used in setting cabin
noise limits may increase with increased blade count. This increase can be especially
important during high-speed cruise, when thickness noise is an important source.

Propeller Diameter
Increasing propeller diameter reduces the blade loading. Thus, for a given thrust
requirement the loading per unit area is reduced, with a corresponding reduction
in loading noise. Increasing diameter is thus beneficial in reducing noise during
takeoff. In addition, at low speed, propellers tend to be more efficient with increased
diameter. Therefore, for a given thrust requirement less power is required, with less
energy put into the system. Increased diameter can be combined with reduced tip
speed for even more noise reduction.

Blade Shape
The effect on noise of blade design parameters such as twist and planform
distributions is more difficult to determine by inspection because they change
aerodynamic loading distribution. Although this can be done by parametric variation
using a noise calculation procedure, most studies show that the noise reduction
potential is small. The actual reduction to be realized depends on the starting point,
but for reasonable designs the potential seems to be about 3 dB. This reduction can
be realized with varying amounts of aerodynamic performance loss. The effect of
blade design has a stronger impact on aerodynamic performance than on noise.

Airfoil Section
Some airfoil sections appear better for noise reduction than others. In general,
however, the airfoil shape has only a small effect in the lower harmonics. Only for
propfans at high speed do the airfoil shape effects appear at the lower harmonics.
The reader is cautioned that the foregoing discussion should be applied only
in the context of a complete system study. Generally, the best approach requires
a complete aerodynamic and acoustic methodology so that the trade-offs between
noise and performance can be evaluated. Other factors such as weight, cost, and
reliability must also to be considered.

Synchrophasing
Synchrophasing is not a means for reducing noise at the source, but rather it
relies on phasing two or more sources to promote noise cancellation. This is done by

60
 Propeller and Propfan Noise

phasing the rotation position of the blades on opposite sid he fuselage so that
the sound impinging on the fuselage has a certain phase re1 ip which promotes
noise cancellation within the cabin. The process by which the noise cancels is too
complex to define analytically to the degree sufficient to realize a reduction. All
implementation of noise reduction by synchrophasing has been done experimentally
and applied to cabin noise (refs. 74 and 75). Reductions of up to 15 dB may be
obtained under specific conditions in limited areas of an airplane cabin, but general
reductions of maximum noise throughout the cabin are less.

References
1. Mikkelson, Daniel C.; Mitchell, Glenn A.; and Bober, Lawrence J.: Summary of Recent NASA
Propeller Research. Aerodynamics and Acoustics of Propellers, AGARD-CP-366, Feb. 1985,
pp. 12-1-12-24. (Available from DTIC as AD A156 047; also available as NASA TM-83733.)
2. Magliozzi, B.: Noise Characteristics of a Model Counterrotating Prop-Fan. AIAA-87-2656,
Oct. 1987.
3. Dobrzynski, Werner M.; Heller, Hanno H.; Powers, John 0.; and Densmore, James E.:
DFVLR/FAA Propeller Noise Tests in the German-Dutch Wind Funnel DNW. DFVLR-IB-
129-86/3, FAA-AEE 86-3, 1986.
4. Ffowcs Williams, J. E.; and Hawkings, D. L.: Sound Generation by Turbulence and Surfaces
in Arbitrary Motion. Philos. Trans. Royal SOC.London, ser. A, vol. 264, no. 1151, May 8,
1969, pp. 321-342.
5. Farassat, F.: Linear Acoustic Formulas for Calculation of Rotating Blade Noise. A I A A J.,
vol. 19, no. 9, Sept. 1981, pp. 1122-1130.
6. Morse, Philip M.; and Ingard, K. Uno: Theoretical Acoustics. McGraw-Hill Book Co., c.1968.
7. Goldstein, Marvin E.: Aeroacoustics. McGraw-Hill Book Co., c.1976.
8. Ashley, Holt; and Landahl, Marten: Aerodynamics of Wings and Bodies. Addison-Wesley
Publ. Co., Inc., c.1965.
9. Zorumski, William E.; and Weir, Donald S., eds.: Aircraft Noise Prediction Program Theoret-
ical Manual-Propeller Aerodynamics and Nozse. NASA TM-83199, Pt. 3, 1986.
10. Farassat, F.; Padula, S. L.; and Dunn, M. H.: Advanced Turboprop Noise Prediction Based on
Recent Theoretical Results. J. Sound €4 Vibration, vol. 119, no. 1, Nov. 22, 1987, pp. 53-79.
11. Hanson, Donald B.: Near Field Noise of High Tip Speed Propellers in Forward Flight. AIAA
Paper No. 76-565, July 1976.
12. Farassat, F.: Advanced Theoretical Treatment of Propeller Noise. Propeller Performance and
Noise, VKI-LS 1982-08, Volume 1, Von Karman Inst. of Fluid Dynamics, May 1982.
13. Farassat, F.; and Succi, G. P.: The Prediction of Helicopter Rotor Discrete Frequency Noise.
Vertica, vol. 7, no. 4, 1983, pp. 309-320.
14. Farassat, F.: Theoretical Analysis of Linearized Acoustics and Aerodynamics of Advanced
Supersonic Propellers. Aerodynamics and Acoustics of Propellers, AGARD-CP-366, Feb. 1985,
pp. 10-1-10-15. (Available from DTIC as AD A156 047.)
15. Gutin, L.: On the Sound Field of a Rotating Propeller. NACA TM 1195, 1948.
16. Hanson, Donald B.: Helicoidal Surface Theory for Harmonic Noise of Propellers in the Far
Field. A I A A J., vol. 18, no. 10, Oct. 1980, pp. 1213-1219.
17. Hanson, Donald B.: Influence of Propeller Design Parameters on Far Field Harmonic Noise in
Forward Flight. A I A A J., vol. 18, no. 11, Nov. 1980, pp. 1313-1319.
18. Hanson, Donald B.: Near-Field Frequency-Domain Theory for Propeller Noise. A I A A J.,
vol. 23, no. 4, Apr. 1985, pp. 499-504.
19. Hanson, D. B.: Shielding of Prop-Fan Cabin Noise by the Fuselage Boundary Layer. J . Sound
& Vibration, vol. 92, no. 4,Feb. 22, 1984, pp. 591-598.
20. Lowson, M. V.: Theoretical Studies of Compressor Noise. NASA CR-1287, 1969.

61
Magliozzi, Hanson, and Amiet

21. Wright, S. E.: Discrete Radiation From Rotating Periodic Sources. J. Sound €4 Vibration,
vol. 17, no. 4, Aug. 22, 1971, pp. 437-498.
22. Hanson, Donald B.: Noise of Counter-Rotation Propellers. J. Aircr., vol. 22, no. 7, July
1985, pp. 609417.
23. Hanson, D. B.; and Fink, M. R.: The Importance of Quadrupole Sources in Prediction of
Transonic Tip Speed Propeller Noise. J. Sound €4 Vibration, vol. 62, no. 1, Jan. 8, 1979,
pp. 19-38.
24. Blackburn, H. W.: Quadrupoles in Potential Flow: Two Model Problems. J. Fluid Mech.,
V O ~ .116, Mar. 1982, pp. 507-530.
25. Curle, N.: The Influence of Solid Boundaries Upon Aerodynamic Sound. Proc. Royal SOC.
(London), ser. A, vol. 231, no. 1187, Sept. 20, 1955, pp. 505-514.
26. Ffowcs Williams, J. E.; and Hall, L. H.: Aerodynamic Sound Generation by Turbulent Flow in
the Vicinity of a Scattering Half Plane. J. Fluid Mech., vol. 40, pt. 4, Mar. 9, 1970, pp. 657470.
27. Chase, David M.: Sound Radiated by Turbulent Flow Off a Rigid Half-Plane as Obtained
From a Wavevector Spectrum of Hydrodynamic Pressure. J. Acoust. SOC.America, vol. 52,
no. 3, pt. 2, Sept. 1972, pp. 1011-1023.
28. Amiet, R. K.: Noise Due to Turbulent Flow Past a Trailing Edge. J. Sound €4 vibration,
vol. 47, no. 3, 1976, pp. 387-393.
29. Amiet, R. K.: Effect of the Incident Surface Pressure Field on Noise Due to Turbulent Flow
Past a Trailing Edge. J. Sound B Vibration, vol. 57, no. 2, 1978, pp. 305-306.
30. Corcos, G. M.: The Structure of the Turbulent Pressure Field in Boundary-Layer Flows.
J. Fluid Mech., vol. 18, pt. 3, Mar. 1964, pp. 353-378.
31. Willmarth, W. W.; and RON, F. W.: Resolution and Structure of the Wall Pressure Field
Beneath a Turbulent Boundary Layer. J. Fluid Mech., vol. 22, pt. 1, May 1965, pp. 81-94.
32. Eckert, E. R. G.; and Drake, Robert M., Jr.: Heat and Mass Transfer, Second ed. McGraw-Hill
Book Co., Inc., 1959.
33. Yu, J. C.; and Joshi, M. C.: On Sound Radiation From the Trailing Edge of an Isolated Airfoil
in a Uniform Flow. AIAA Paper No. 79-0603, Mar. 1979.
34. Schlinker, Robert H.; and Amiet, Roy K.: Helicopter Rotor Trailing Edge Noise. NASA
CR-3470, 1981.
35. Brooks, T. F.; and Hodgson, T. H.: Prediction and Comparison of Trailing Edge Noise Using
Measured Surface Pressures. AIAA-80-0977, June 1980.
36. Chou, S.-T.; and George, A. R.: Effect of Angle of Attack on Rotor Trailing-Edge Noise.
A I A A J., vol. 22, no. 12, Dec. 1984, pp. 1821-1823.
37. Fink, Martin R.: Noise Component Method for Airframe Noise. AIAA Paper 77-1271,
Oct. 1977.
38. Roy, D.: Doppler Frequency Effects Due to Source, Medium, and Receiver Motions of Constant
Velocity. AIAA-83-0702, Apr. 1983.
39. Kinder, Lawrence E.; and Frey, Austin R.: Fundamentals of Acoustics, Second ed. John Wiley
& Sons, Inc., c.1962.
40. Noise Standards: Aircraft Type and Airworthiness Certzfication. FAR, Pt. 36, Federal Aviation
Adm., June 1974. (Consolidated reprint Aug. 12, 1985.)
41. McAninch, G. L.; and Rawls, J. W., Jr.: Effects of Boundary Layer Refraction and Fuselage
Scattering on Fuselage Surface Noise From Advanced Turboprop Propellers. AIAA-840249,
Jan. 1984.
42. Hanson, D. B.; and Magliozzi, B.: Propagation of Propeller Tone Noise Through a Fuselage
Boundary Layer. J. Aircr., vol. 22, no. 1, Jan. 1985, pp. 63-70.
43. Amiet, R. K.: Diffraction of Sound by a Half-Plane in a Uniform Flow. UTRC Rep. 86-17,
United Technologies Research Center, Aug. 1986.
44. American National Standard Method for the Calculation of the Absorption of Sound by the
Atmosphere. ANSI SI.26-1978 (ASA 23-1978), American Inst. of Physics, 1978.

62
 Propeller and Propfan Noise

45. Sutherland, Louis C.: Development of Improved Methods for Computing Air Absorption Losses.
Volume I: Evaluation of Vibrational Relaxation Frequencies From Laboratory Data. WR 86-22
(Contract DTFA01-84C-00003), Wyle Research, Feb. 17, 1987.
46. Marsh, Alan H.: Evaluation of Alternative Procedures for Atmospheric Absorption Adjustments
During Noise Certification. FAA-EE80-46, Oct. 1980.
Volume I: Analyses and Results.
Volume 11: Computer Program.
Volume 111: Tables of Atmospheric-Absorption Losses.
47. Standard Values of Atmospheric Absorption as a Function of Temperature and Humidity.
ARP 866A, SOC.of Automotive Engineers, Mar. 15, 1975.
48. Acoustic Effects Produced by a Reflecting Plane. AIR 1327, SOC.of Automotive Engineers,
Inc., Jan. 15, 1976.
49. Practical Methods To Obtain Free-Field Sound Pressure Levels From Acoustical Measurements
Over Ground Surfaces. AIR 1672B, SOC.of Automotive Engineers, Inc., June 1983.
50. Yoerkie, C. A.; and Larson, R. S.: Prediciton of ]Free-Field Noise Levels From Pole Microphone
Measurements. AIAA-80-1058, June 1980.
51. Pao, S. Paul; Wenzel, Alan R.; and Oncley, Paul B.: Prediction of Ground Effects on Aircraft
Noise. NASA TP-1104, 1978.
52. Zorumski, William E.: Prediction of Aircraft Sideline Noise Attenuation. NASA TM-78717,
1978.
53. Chessell, C. I.: Propagation of Noise Along a Finite Impedance Boundary. J. Acoust. SOC.
America, vol. 62, no. 4, Oct. 1977, pp. 825-834.
54. Estimation of One-Third-Octave-Band Lateral Attenuation of Sound From Jet-propelled Air-
planes. AIR 1906, SOC.of Automotive Engineers, Inc., Oct. 1985.
55. Hawkings, D. L.; and Lowson, M. V.: Theory of Open Supersonic Rotor Noise. J. Sound €d
Vibration, vol. 36, no. 1,Sept. 8, 1974, pp. 1-20.
56. Barger, Raymond L.: Theoretical Prediction of Nonlinear Propagation Effect on Noise Signa-
tures Generated by Subsonic or Supersonic Propeller- or Rotor-Blade Tips. NASA TP-1660,
1980.
57. Tam, Christopher K. W.; and Salikuddin, M.: Weakly Nonlinear Acoustic and Shock-Wave
Theory of the Noise of Advanced High-speed Turbopropellers. J. Fluid Mech., vol. 164,
Mar. 1986, pp. 127-154.
58. Lindblad, Ingemar A. A.: Asymptotic Analysis of the Transonic Region of a High-speed
Propeller. AIAA-89-1077, Apr. 1989.
59. Magliozzi, B.: The Influence of Forward Flight on Propeller Noise. NASA CR-145105, 1977.
60. Pegg, R. J.; Farassat, F.; and Magliozzi, B.: Some Measured and Calculated Effects of Forward
Velocity on Propeller Noise. ASME Paper 77-GT-70, Mar. 1977.
61. Hubbard, Harvey H.: Sound From Dual-Rotating and Multiple Single-Rotating Propellers.
NASA TN 1654, 1948.
62. Glover, B. M., Jr.; Plunkett, E. I.; and Simcox, C. D.: Noise Testing of an Advanced Design
Propeller in the Boeing Transonic Wind Tunnel With and Without Test Section Acoustic
Treatment. AIAA-842366, Oct. 1984.
63. Amiet, Roy K.: Correction of Open Jet Wind Tunnel Measurements for Shear Layer Refraction.
AIAA Paper 75-532, Mar. 1975.
64. Ross, R.; Young, K. J.; Allen, R. M.; and Van Ditshuizen, J. C . A.: Acoustic Wave Propagation
Through the Shear Layer of the DNW Large Open Jet Wind Tunnel. AIAA-83-0699, Apr. 1983.
65. Herkes, W. H.; Strout, F. G.; and Ross, R.: Acoustic Evaluation of DNW Free Jet Shear Layer
Correction Using a Model Jet. AIAA-83-0757, Apr. 1983.
66. Brooks, Bennett M.; and Mackall, Karen 6.: Measurement and Analysis of Acoustic Flight
Test Data for Two Advanced Design High Speed Propeller Models. AIAA-84-0250, Jan. 1984.
67. Weir, D. S.; and Powers, 3. 0.: Comparisons of Predicted Propeller Noise With Windtunnel
and Flyover Data. AIAA-87-0527, Jan. 1987.
Magliozzi, Hanson, and Amiet

68. Succi, George P.; Munro, David H.; and Zimmer, Jeffrey A.: Experimental Verification of
Propeller Noise Prediction. AIAA-80-0994, June 1980.
69. Nallasamy, M.; Woodward, R. P.; and Groeneweg, J. F.: High Speed Propeller Performance
and Noise Predictions at Takeoff/Landing Conditions. AIAA-88-0264, Jan. 1988.
70. Magliozzi, B.: V/STOL Rotary Propulsor Noise Prediction Model-Ground Reflection Effects
and Propeller Thickness Noise. Rep. No. FAA-EE82-15, Aug. 1982.
71. Environmental Protection-Annex 16 to the Convention on International Civil Aviation.
Volume I-Aircraft Noise, First ed. International Civil Aviation Organization, Nov. 26, 1981.
(As amended Nov. 21, 1985.)
72. Parzych, D. J.; Magliozzi, B.; and Metzger, F. B.: Prop-Fan/Turboprop Acoustic Terminology.
SAE Tech. Paper Ser. 871839, Oct. 1987.
73. Prediction Procedure for Near-Field and Far-Field Propeller Noise. AIR 1407, SOC. of
Automotive Engineers, Inc., May 1977.
74. Johnston, J. F.; Donham, R. E.; and Guinn, W. A.: Propeller Signatures and Their Use.
AIAA-80-1035, June 1980.
75. Magliozzi, B.: Synchrophasing for Cabin Noise Reduction of Propeller-Driven Airplanes.
AIAA-83-0717, Apr. 1983.

64