You are on page 1of 12

Propeller Noise

Rod H. Self
Institute of Sound and Vibration Research, University of Southampton, Southampton, UK

1 Introduction 2 Propeller Tonal Noise Prediction 3 Propeller Broadband Noise Prediction 4 Installed Propeller Noise Prediction 5 Propeller Noise Reduction 6 Perspective Acknowledgments Related Chapters References

1 5 8 8 9 10 10 10 10

1 INTRODUCTION
Propellers were the rst means of powering aircraft and propeller technology advanced steadily until the late 1950s. At this point, the advantages of the newly developed jet engine allowed it to become the dominant technology. The most important of these advantages was that jet-powered aircraft could y much faster something that propellers of the time could not match. Consequently, the majority of the commercial eet is now powered by turbofan engines. However, propellers have never disappeared and are still widely used for lightweight aircraft and small commuter aircraft. An important advantage of propellers over turbofans is their much higher installed efciency leading to reduced fuel consumption. This led to a minor renaissance in the 1970s and 1980s when higher oil prices prompted manufacturers to look at more efcient propulsion technologies for aircraft,
Encyclopedia of Aerospace Engineering. Edited by Richard Blockley and Wei Shyy c 2010 John Wiley & Sons, Ltd. ISBN: 978-0-470-68665-2

and several designs of prop-fan (high-speed turboprop otherwise known as unducted fans or advanced open rotor) were built and tested. Many of the earlier structural problems were overcome by better design aided by advances in computational aerodynamics and structural mechanics and by the use of new composite materials. These advanced prop-fans were certainly efcient at higher Mach numbers, but, for a number of reasons, they were never used for commercial aircraft. An often-cited disadvantage is the noise produced by prop-fans compared to turbofans, but perhaps a more realistic assessment is that the subsequent fall in oil prices removed the incentive to invest in overcoming such problems. This is not something that can be said today. Oil prices are again seeing upward pressure added to which there are concerns over emissions in the form of greenhouse gases (CO2 ) and local pollutants such as NOX . Thus, a need to nd substantially more efcient aircraft has returned. Advanced prop-fans are again being investigated seriously by both engine manufacturers and airframers as a possible means of powering short- and medium-haul aircraft. Methods of predicting propeller noise have been under development for 85 years. But just as the interest in propellers from an aerodynamic standpoint has waxed and waned, so there have been bouts of activity in noise research interspersed with periods of relative inactivity. As with many aeroacoustic problems, theoretical deciencies and a lack of computational power meant that empirical methods were the most reliable early on, and these have continued to be developed, as have the theoretical methods of noise prediction. Gutin (1936) developed the rst theory of propeller noise, and this has been much developed by subsequent investigators. The interest in prop-fans during the 1980s resulted in a number of theoretical advances including the ability to account for high-tip Mach numbers and the additional sources arising

2 Acoustics and Noise with counter-rotating propellers (CRPs). Today, advances in computational uid dynamics are allowing researchers to better address the effects of installation on propeller noise due to non-uniform ows. It is not possible in a short chapter such as this to give an exhaustive account of the subject. Rather, the intention is to give the reader an overall avor and point him or her to more substantive accounts. The chapter is organized as follows. The remaining parts of the introduction describe various types of propeller and give a qualitatively description of propeller noise characteristics and sources. Methods of predicting tone noise are discussed in Section 2 together with methods for broadband noise prediction in Section 4. It is important to realize that an installed propeller will have signicantly different noise levels and characteristics than the corresponding isolated propeller. This is discussed in Section 5. The nal section gives some methods for reducing propeller noise.

Figure 2. Modern conception of a counter-rotating propeller engine. This engine differs from those shown in Figure 1 by being designed as an-aft mounted pusher conguration. Image courtesy of c Rolls-Royce.

1.1 Description of propeller types


Propellers are instantly recognizable and can be described as an unshrouded (i.e., open), rotating set (or sets) of blades. While there are many variations on the basic theme (principally number and shape of blades), they all operate in the same way. Each blade acts as an aerofoil with high and low pressure surfaces that induces an airow, thus creating thrust. For low ight speeds the thrust is provided by a relatively large amount of air moving quite slowing and the propeller is highly efcient. However, as the ight speed increases and the efciency decreases. To counter this, prop-fans were introduced. Prop-fans are characterized by a larger number of smaller diameter blades and increased rotation speed. Additionally, the blades are often swept and a second, counter-rotating set of blades may be added behind the rst. Examples of each of these propeller types are shown in Figure 1. The typical high-performance propeller as used on commuter airplanes is shown in Figure 1a. These propellers have two to six relatively straight, narrow blades and are highly

efcient at lower ight speeds (Mach number approximately 0.6), but this efciency falls off rapidly as the ight speed increases. For these increased speeds, the prop-fan illustrated in Figure 1b is more suitable. The prop-fan has a higher number of thin, highly swept blades. This allows it to operate at higher loadings for the same thrust and allows the diameter of the propeller to be reduced. The counter-rotating propeller illustrated in 1c consists of two coaxial propellers rotating in opposite directions. The upstream propeller produces a swirling wake that increases the air velocity onto the downstream propeller blades increasing the thrust produced by the downstream blade row (relative to an identical propeller in a single rotation conguration). The swirl is then eliminated, or at least signicantly reduced, by the downstream propeller that rotates in the opposite direction. This reduction of the swirl in the propeller wake can signicantly increase fuel efciency relative to an equivalent single rotation propeller design. A modern conception of a CRP is illustrated in Figure 2. Notice that in the conguration shown, the propellers sit at the rear of the turbine housing. For this reason, it is known as a pusher conguration.

1.2 Propeller noise characteristics


Propellers produce two distinctly different types of noise. The rst of these is tonal or harmonic noise coming from sources that repeat exactly during each revolution of the propeller. The second is broadband noise that is a random non-periodic signal caused by turbulent ow over the propeller blades. For an ideal propeller with B blades rotating with shaft rotation angular frequency rad s1 , the fundamental component of the

(a)

(b)

(c)

Figure 1. (a) Conventional propeller; (b) prop-fan; (c) counterrotating propeller. Reproduced with permission from Hubbard (1995).

Propeller Noise 3 Unsteady sources arise principally from the fact that propellers rarely operate in a completely clean inow. For example, when a propeller operates with its axis at an angle to the incoming airow, the loading on the blades varies periodically as it rotates. Such periodic changes in loading can also occur when the propeller is installed in close proximity to the aircraft fuselage or wing. As the propeller rotates through the varying ow eld, the loading on the blade varies accordingly, producing tones at harmonics of the blade passing frequency. These so-called distortion tones add or subtract from the steady loading noise in a highly directional way complicating the prediction of the noise eld considerably. Unsteady loadings are important sources of noise for counter-rotating prop-fans. This is because each set of blades distorts the ow eld seen by the other set of blades. There are three principal causes. Tipvortex interaction tones arise when the tip vortices produced by each upstream propeller blade convect downstream and interact with the downstream propeller blades. (This source is also found in helicopter rotors where it is known as blade vortex interaction, BVI, giving rise to the characteristic thumping sound.) A numerical simulation of the pressure elds associated with CRP blades is shown in Figure 4, where the tip vortices can be clearly seen. Viscous wake interaction tones arise from the unsteady lifting on blades in the second row when they pass through the viscous wakes shed from the rst row. Bound potential interaction tones are caused by the bound potential eld, due to the thickness and loading on each propeller blade, interacting with the blades on the adjacent propeller (Figure 5).

Figure 3. Conventional propeller noise spectrum.

tonal noise will be at the blade passing frequency B/2 Hz and harmonics thereof. By contrast, the broadband noise is random in nature and contains components at all frequencies. Generally, the tonal noise components dominate at lower frequencies with the broadband noise increasing in importance as the frequency increases. The relative importance of two types of noise can be gauged from Figure 3 that illustrates a typical noise spectrum for a conventional propeller. Broadband noise sources are discussed in Section 3 below.

1.3 Overview of propeller tonal noise sources


Sources of propeller tonal noise are often classied as being steady or unsteady. Essentially, steady sources are those that are constant when viewed by an observer using a reference frame that is xed to the rotating blades while unsteady sources exhibit a time dependence to an observer using the rotating reference frame. The steady sources give rise to rotor-alone tones that are produced by the rotating steady loading and thickness of each blade. In the case of the steady loading source, the lift and drag on the blade are constant in the rotating frame, but, when viewed from the rotating frame, the components of these forces are varying in time giving rise to a time-varying pressure eld that propagates as noise. Thickness noise arises from the time-varying displacement of air by the blade volume as it rotates in the xed frame. It is purely dependent on the geometry and operating condition of the engine.The simplest way of reducing thickness noise is to decrease the blade volume (especially near the faster moving tip). If the blade thickness is reduced by half then thickness noise is reduced by 6 dB.

Figure 4. Numerical simulation of pressure contours for a counterrotating propeller showing the tip vortices. Image courtesy of Alexios Zachariadis, University of Cambridge.

4 Acoustics and Noise In all cases, the generated noise is tonal. If the two rotors have the same number of blades and are operating at the same rotational speed, then the unsteady sources produce tones at the same frequencies as the steady sources that are the harmonics of the blade passing frequency. Generally, however, the two rotors have different numbers of blades and rotate at different shaft frequencies. In this case, the interaction tones are quite distinct, and the number of individual frequencies at which they occur is large. These interaction tones will be the dominant noise sources at low-speed operations such as takeoff but will be less important at high-speed cruise conditions where the steady sources become dominant. Whatever their source, the tones produced are highly directional in nature. That is, the level of noise varies as the observer changes angular position relative to the propeller axis (Figure 6). Notice that different tones have different directivities, with interaction tones having a more complicated directivity than rotor-alone tones.

Figure 5. Bound potential eld of the rst blade row interacting with the blades on the second row to produce noise.

150

140

130

120 Sound pressure level (dB)

110

100

90

80

70

60

50

20

40

60

80

100

120

140

160

180

Observer angle ()
Viscous wake Rearward potential Forward potential Tip vortex Total

Figure 6. Directivity of interaction tones. The graphs show how the levels of noise vary for a number of different interaction tones as the observer changes angle to the propeller axis. The directivity of steady tones is somewhat simpler.

Propeller Noise 5

2 PROPELLER TONAL NOISE PREDICTION


The rst complete description of propeller steady loading noise was given by Gutin (1936), but modern methods all start with the equation of Ffowcs-Williams and Hawkins (1969). This is an extension of Lighthills (1952) theory and Curles (1955) work on solid boundaries to include the effects of motion: 2 1 2 2 2 2 c t xi p= 2 T ij + xi xj xi t 0 vi (f ) pij (f ) f xi f xj (1)

sources are represented as a series of rotating lines. Liftingline methods are reectively easy to code but are limited in application blades with small chords. A more realistic model is given by lifting-surface methods. Here, the blades are assumed to be thin with small angles of attack but are no longer assumed to be chordwise compact. The sources are now seen as located on an innitely thin multiple helix described by the blades. Lifting-surface methods have been developed in both the time domain (Farassat, 1986) and frequency domain (Hanson, 1980).

2.1 Time domain methods


Equation (1) or (2) may be solved directly in terms of integrals over the blade surface with time dependency. These time domain methods allow the blade geometry to be treated with great precision but suffer the disadvantage that the integrands need to be computed at retarded times. That is, to compute the acoustic pressure p(xi , t) at the observer position xi and time t, it is necessary to determine where every element of the surface was, and the source strength, when the element emitted a wave that arrives at the observer position xi at time t. Additionally, a direct integration of either form of the Ffowcs WilliamsHawkins equation requires the numerical computation of derivatives. Time domain methods require great accuracy and are difcult to code. However, despite the numerous complications, a number of useful time domain codes have been developed, most notably by Farassat who has developed a number of formulations. Codes that incorporate the work of Farassat include the Aircraft Noise Prediction Program (Zorumski and Weir, 1986), the DunnFarassatPadula Advanced Turboprop Prediction (Farassat, Padula and Dunn, 1987), and the Advanced Subsonic and Supersonic Propeller Induced Noise (ASSPIN) prediction code (Dunn and Tarkenton, 1992).

The left-hand side of this equation is the usual linear wave operator operating on the acoustic pressure, p, while the righthand side shows that there are three sources of sound. The rst term represents a distribution of quadrupoles of strength given by the Lighthill stress tensor Tij outside of any surface and zero within them (indicated by the overline). The second term includes the compressive stress tensor pij and represents the action of the blade forces on the air, which give rise to loading noise. The blade surface is given by f = 0, represents the Dirac delta function, and vi are the components of the local surface velocity. The third term quanties the effects of the blade separating the air as it rotates producing thickness noise. It is usual to simplify this equation in a way that eliminates the need to consider the details of the blades cross-sectional geometry as represented by the function f. This will not lead to signicant error unless very high frequencies (those with wavelength comparable to the blade thickness) are considered. In this case, the quadrupole source is also ignored leading to a linear equation of the form 1 2 2 2 2 2 c t xi p = 0 Q+ Fi t xi (2)

2.2 Frequency domain methods


Frequency domain methods obtained by taking a Fourier transformation over time t are popular tools as they eliminate the need for computing retarded time blade locations and numerical derivatives. This makes them easier to code and decreases the computing power required but does so at the expense of accurately representing the blade geometry but this is unlikely to be a problem unless results to a high number of harmonics are required. The main proponent of frequency domain methods is Hanson who has developed formulae to account for a number of different sources and effects. A useful introductory

For details of the derivation, see, for example, Morse and Ingard (1968) or Goldstein (1976). In this simplied equation, the thickness noise source is now represented as an equivalent distribution of volume sources (and sinks), Q, while determination the loading sources reduces to nding the form of the force Fi imposed by the blade on the air. There are two common approximations used to tackle these equations. Lifting-line methods utilize an asymptotic expansion with the reciprocal of the blade aspect ratio as small parameter (Brouwer, 1992). In this case, variations in the chordwise direction are collapsed to a single point, and the

6 Acoustics and Noise summary to the main elements of Hansons work can be found in Chapter 1 of Hubbard (1995). Hansons original work has been extended and applied to counter-rotating propellers (Hanson, 1985; Parry, 1988). Codes using the frequency domain approach include Whiteld et al. (1989) and Engineering Sciences Data Unit (ESDU) (1996) for single propellers and Whiteld, Mani and Glieb (1990a, 1990b) for counter-rotating propellers. coordinate z = r/rt where rt is the tip radius. Blade sweep is accounted for by the mid-chord alignment, MCA, which is a measure of how far the mid point of the chord is offset from the radius vector (see Figure 7). To determine the thickness noise requires that the thickness of the blade is specied at each radial location. This specication is given as h(X)T where h(X) is a non-dimensional thickness distribution function ranging from 0 at each edge of the blade to 1 at the thickest part of the section where the thickness is T. Similarly, to determine the amplitudes arising from lift and drag requires analogous distributions to be dened. These are specied in terms of section lift and drag coefcients CL and CD and chordwise lift and drag distribution functions fL (X) and fD (X). CL and CD are dened so that the lift and drag force per unit span-wise distance is given by 1 0 Ur2 CL,D 2 (5)

2.2.1

Steady sources

Hansons method is a lifting-surface method, which we illustrate by considering the steady thickness and loading sources. In the far-eld, the pressure is given as an innite sum over harmonics of the blade passing frequency:

p(t) =
m=

PBm exp(imB t)

(3)

where the propeller has B blades rotating at an angular frequency and the Fourier amplitudes PBm can be written as a sum PBm = PVm + PLm + PDm (4)

where Ur is the blade section speed given by Ur = c0 Mr where Mr is the section relative Mach number, Mr =
2 MX + z2 Mt2

(6)

of amplitudes arising from the blades thickness, lift, and drag. These are determined by integration over a blade planform dened in terms of a normalized axial coordinate X = x/b where b is the local blade chord and a normalized radial

The equivalent of a retarded time appears in the frequency domain formulation via a retarded radiation angle (also called emission angle) that is related to the current (polar) angle

Figure 7. Denition of the mid-chord alignment MCA, and the coordinate system for the propeller disk with observer in the xt plane.

Propeller Noise 7 between aircraft and observer, t , by cos = cos t 1


2 MX

2.2.2
2

Unsteady sources

sin t + MX sin t
2

(7)

Frequency domain methods for unsteady sources can become quite cumbersome and will not be given in detail. Essentially unsteady loading on the blades is dealt with by expanding the lift and drag coefcients as Fourier series such as

and Hanson denes the non-dimensional wave numbers 2mBbMt kX = D(1 MX cos )Mr and ky = 2mBb(MX Mr2 cos ) D(1 MX cos )zMr (8) CL =
j=

CLj eij t

(13)

(9)

where D = 2rt is the blade diameter. Using these, the thickness and loading distributions are given in terms of their Fourier transforms for example,
+1/2 1/2

V (kX )

h(X)eikX X dX

(10)

with similar expressions dening L (kX ) and D (kX ) as transforms of fL (X) and fD (X). The nal expression for the thickness noise is now given by
2 0 c0 sin exp imB r c0

PVm =

8 y(1 MX cos )/D


1 0

Mr2 exp(is )JmB


V (kX )

mBzMt sin 1 MX cos (11)

In this case, the j = 0 lifting coefcient is the steady, or mean, loading previously considered. It is possible to obtain formulations for a general case of harmonic loading at any frequency (Hanson, 1985), but two special cases are worthy of note. When the propeller is operating at an angle to an otherwise undistorted inow, the fundamental frequency in (13) will be the shaft rotation frequency and j = j . For contra-rotating prop-fans, it is necessary to consider the variations in loading and drag that arise from the ow eld of one set of blades interfering with the other set of blades, and, in this case, the load frequencies will be of the form j = jB1 ( 1 + 2 ). When the drag coefcient is similarly expanded, the result is that the unsteady loading equivalent of equation (11) will contain a double sum, and this has the effect of greatly multiplying the number of tones present in the noise spectrum. In the contra-rotating code developed by Whiteld, Mani and Gliebe (1990a, 1990b), the unsteady loading sources arising from non-uniform inow onto the blades as well as the unsteady loading on the second set of blades due to the viscous wake and tip vortex of the rst set of blades are considered. Parry (1988) derives expressions for unsteady loading due to the bound potential elds of the blades.

2 kX T

2.2.3
dz

Nonlinear effects

where y = R sin is the altitude of the aircraft. In this equation, JmB is a Bessel function of order mB and s is the phase lag arising from the blade sweep and is given by s = MCA 2mBbMt /D D Mr (1 MX cos )

(12)

Expressions for PLm and PDm are obtained from equation (11) by replacing the term in braces in the integral on the right-hand side by ikX CL L (kX )/2 and ikX CD D (kX )/2, respectively.

For lower blade speeds, the steady loading and thickness noise sources together with the unsteady sources will dominate. However, as the blade speed increases and approaches high subsonic values, it is possible for local areas of supersonic ow to exist. This is called transonic ow, and a shock is usually formed on the suction side of the blades. In this case, noise due to the nonlinear quadrupole term in the Ffowcs-WilliamsHawkings equation (1) will make a signicant contribution to the overall sound pressure level. Parry (1988) points out that the incorporation of blade sweep into the propeller design can effectively remove this transonic phenomenon and that, therefore, it should be possible to neglect the quadrupole source. When it is necessary to consider transonic effects, a discussion is given by Brouwer (1994),

8 Acoustics and Noise who also includes formula for the effects of supersonic blade speeds.

4 INSTALLED PROPELLER NOISE PREDICTION


The acoustic formulations discussed in previous sections assume that the sound is radiated into a free eld containing no reecting surfaces and with no other propagation effects other than that of spherical spreading. In practice, the installation of the open rotor means the radiated sound eld is altered by the presence of surfaces such as the aircraft wing and fuselage. The forward motion of the aircraft further modies the noise received by an observer on the ground by inducing a Doppler frequency shift. These effects are usually signicant and applying prediction codes without correctly allowing for them can result in estimates of the noise that substantially underpredict.

3 PROPELLER BROADBAND NOISE PREDICTION


Propeller broadband noise is much smaller than the tonal noise components, and, for this reason, it is often ignored. It arises by way of interaction of the blade with turbulence. Physically, turbulence acts to produce a randomly uctuating force on the propellers edges, and broadband noise is therefore dipole in nature. The turbulence may be generated upstream of the propeller and be convected onto the blades by the mean stream or it may be generated in the turbulent boundary layer of the blade. When turbulence is convected onto the blades, the noise generation is generally concentrated on the leading edge of the blade. This requires that the acoustic wavelength, , is less than the blade chord, b. If the frequency is low enough then the entire blade will radiate (compact case). The principal problem in this case is quantifying the inow turbulence, which is often difcult. For turbulence generated in the boundary layer, the situation is necessarily non-compact and the sound generation is concentrated on the trailing edge of the blade where the turbulent eddies induce changes in the loading on the blade. Also, we are far more likely to be able to make an estimate of the amplitude and spectral characteristics of the turbulent uctuations allowing some general expressions to be derived. One method for predicting trailing-edge noise is based on the work of Amiet (1976), and an example of its implementation is given in Schlinker and Amiet (1981). While such methods allow estimates of the 1/3 octave Sound Pressure Level (SPL) levels to be made, these are highly dependent on the assumed spectral characteristics of the turbulence, which are normally estimated by tting to data. A discussion of this method is provided in Chapter 1 of Hubbard (1995).

4.1 Doppler effects


When an acoustic source is in motion, an observer will hear the sound at a different frequency than the one he would hear it at if the source was not in motion. This effect is known as a Doppler shift and is at its most effective when the source is moving directly towards, or away from, the observer. For aircraft noise, it is therefore the yover noise that is most effected by Doppler shifts (Figure 8). When a source moves in a stationary medium relative to an observer, noise emitted by the source at frequency fs will be observed the at a frequency fo = fs 1 Mx cos (14)

where Mx is the Mach number of the source and is the angle between the source velocity vector and a line joining source and observer at the time the sound is emitted. Generalization of equation (14) to include the case where the medium is also in motion is given by Roy (1983). This

Figure 8. When an aircraft moves towards an observer, the wavefronts are closer together resulting in a higher frequency. When it moves away, the reverse is true and the observer hears a lower frequency.

Propeller Noise 9 allows estimates to be made of the Doppler shift that occurs when wind is present. estimating the shielding of wing installed prop-fans is given by Amiet (1986).

4.2 Refraction effects and shielding and scattering by the airframe


A signicant problem with propeller aircraft is the cabin noise that is caused by vibration of the cabin walls. Some of this vibration is transmitted through the airframe from vibrating components in the engines, but a large part arises from acoustic excitation of the fuselage. The exciting acoustic eld is not the same as the one that would be generated by an isolated propeller in the same location because the boundary layer on the fuselage acts to refract the noise. Sound incident on the fuselage is also scattered and modies the far eld noise, as does sound incident on the wings (Figure 9). To correctly model these phenomena, the neareld radiation must be calculated, and this requires rather more general formula than the ones quoted above. The calculation of the acoustic near-eld and the subsequent scattering by the fuselage is quite complicated and will not be discussed in detail. The problem has been addressed by a number of previous investigations including that of Hanson and Magliozzi (1985) and Whiteld et al. (1989) who describe a theoretical method for calculating the scattering of the sound eld produced by a contra-rotating propeller by the fuselage. Their method models the effect of the boundary layer on the fuselage on the radiated sound eld, which they show has a signicant effect on the radiated sound levels. As well as scattering the sound by the fuselage, wings can be used to provide shielding. This will be most efcient for higher frequencies (i.e., those within or approaching the geometric acoustic asymptotic limit) when the wing blocks the line of sight between observer and source. A method for

5 PROPELLER NOISE REDUCTION


There are a number of reasons that make it desirable to reduce propeller noise as far as possible. First, consideration must be given to possible fatigue of the airframe resulting from the pressure loadings imposed by the acoustic eld, and this is particularly important if the airframe structure has resonances that coincide with one of the propeller frequencies. A second consideration is that of cabin noise that needs to be minimized for passenger and crew comfort. Both of these considerations depend on the acoustic eld in close proximity to the propellers. Also, they both operate during the entire duration of the ight and are therefore most inuenced by the noise during high-speed cruise operating conditions. The third consideration is that of far-eld community noise levels and aircraft certication requirements. A number of guiding principles to reduce propeller noise can be inferred through inspection of the prediction theories.

5.1 Blade number


Probably the best known method of reducing propeller noise is to increase the number of blades. For a given thrust requirement, increasing the number of blades reduces the loading noise, and this is particularly benecial at takeoff where the low-speed operation means that loading noise dominates. However, increasing the number of blades also has negative effects. First, increasing blade number raises the frequencies and this can be detrimental on an A-weighted-based scale such as EPNdB. Second, without altering the volume of the blades, an increase in their number will lead to an increase in thickness noise, and this may have to be offset by altering other parameters.

5.2 Blade thickness


Thickness noise can be signicant during high-speed cruise, when blade relative velocities are high. As thickness noise scales with the blade volume, it can be reduced by reducing either the thickness or the chord, and this implies that narrow, thin blades are optimal for such a reduction. The main limiting factor in achieving this is the structural integrity of the blades. While the overall sound level is dependent on the blade volume, the spectrum is dependent on the shape of the blade so any reductions will depend on the frequency being considered.

Figure 9. Sound generated by propellers can scatter and reect from wings and fuselage. Image courtesy of c Flybe.com.

10 Acoustics and Noise

5.3 Operating conditions


The sound generated by a blade section from all sources depends strongly on the relative velocity, and this is why sound generation is normally maximum at the tip. Consequently, a noise reduction can be made by reducing tip speed. In order to maintain thrust, a compensating increase in blade chord and/or number of blades must be made.

5.4 Blade sweep


Increasing blade sweep is benecial during high-speed cruise, when blade-section-relative Mach numbers are relatively high. Benets during takeoff are considerably less.

5.5 Propeller diameter


An increase in the propeller diameter necessarily entails a reduction in the blade loading. For a given thrust, this implies that the loading per unit area is reduced and hence a reduction in loading noise. This will be benecial at low-speed operation at takeoff where loading noise is important. Signicantly, this also increases the efciency of the propeller allowing a second gain to be made by decreasing the tip speed.

cally propeller driven. Airframe noise will be more important on the proposed larger aircraft that use contra-rotating propellers but there are also a new range of tonal sources. These arise from a range of interactions that occur between the two blade rows. Once again, the theory is well understood for the dominant sources and is beginning to be understood for the other sources. Turbo-fan engines offer considerable scope for noise reduction technology such as liners (see Propulsion System Noise: Turbomachinery) that are not available for propellers. Currently the major strategy for reducing propeller noise is by careful design. This is particularly relevant for the newer generation of CRPs where it is already known that making fairly simple design changes can radically affect the noise. For example, cropping the second set of blades can considerably reduce, if not eliminate, the tip-vortex noise. However, to make this a practical proposition it requires understanding the balance between aerodynamic performance and noise source mechanisms and this is the subject of current research.

ACKNOWLEDGMENTS
The author gratefully acknowledges the assistance of Dr M. Kingan from the ISVR for his insightful comments on the text of this chapter.

5.6 Blade shape and aerofoil cross section


The exact way in which design parameters such as twist or aerofoil cross section affect noise is not immediately obvious. While such factors are known to have an inuence on the noise, this always needs to be considered in relation to the aerodynamic performance of the blade. As such, rules of thumb are of little use, and these factors must be considered as part of an overall optimization and trade-off study.

RELATED CHAPTERS
Ducted and Unducted Fans Commercial Engine Noise

REFERENCES
Amiet, R.K. (1976) Noise due to a turbulent ow past a trailing edge. J. Sound Vib., 47(3), 387393. Amiet, R.K. (1986) Diffraction of sound by a half-plane in a uniform ow. United Technologies Research Center Report No. 86-17. Brouwer, H.H. (1992) On the use of matched asymptotic expansions in propeller aerodynamics and acoustics. J. Fluid Mech., 242, 117142. Brouwer, H.H. (1994) Aircraft propeller noise. NLR Technical Publication TP 94203 U. Curle, N. (1955) The inuence of solid boundaries on aerodynamic sound. Proc. R. Soc., A231, 505514. Dunn, M. and Tarkenton, G. (1992) Computational methods in the prediction of Advanced Subsonic and Supersonic Propeller Induced Noise ASSPIN users manual. NASA CR-4434. ESDU International plc. (1996) Prediction of near-eld and far-eld harmonic noise from subsonic propellers with non-axial inow. ESDU Item No. 95029.

6 PERSPECTIVE
The noise from propeller engines is dominated by tones and a fairly complete theory for these has been developed at least in the case of single propellers. Major contributors are the loading and thickness sources giving rise to rotor-alone tones and distortion tones arising from unsteady ow on to the blades. Solutions are possible in both the time and the frequency domain. Like turbo-fan powered aircraft, turbo-prop aircraft also produce airframe noise (see Airframe Noise: Landing Gear Noise and Airframe Noise: High Lift Device Noise) but these sources are less dominant on the smaller aircraft that are typi-

Propeller Noise 11
Farassat, F. (1986) Prediction of advanced propeller noise in the time domain. AIAA J., 24, 587. Farassat, F., Padula, S.L. and Dunn, M.H. (1987) Advanced turboprop noise prediction based on recent theoretical results. J. Sound Vib., 119(1), 5379. Ffowcs-Williams, J.E. and Hawkings, D.L. (1969) Sound generation by turbulence and surfaces in arbitrary motion. Philos. Trans. R. Soc., A264, 321342. Goldstein, M.E. (1976) Aeroacoustics, McGraw-Hill. Gutin, L. (1936) On the sound eld of a rotating propeller. Zeit. Sowjetunion, 9(1), 5771. (Translated as 1948 NACA Technical Memorandum 1195). Hanson, D.B. (1980) Helicoidal surface theory for harmonic noise of propellers in the far eld. AIAA J., 18(10), 1213 1219. Hanson, D.B. (1985) Noise of counter-rotation propellers. J. Aircraft, 22(7), 609617. Hanson, D.B. and Magliozzi, B. (1985) Propagation of propeller tone noise through a fuselage boundary layer. J. Aircraft, 22(1), 6370. Hubbard, H.H. (ed.) (1995) Aeroacoustics of Flight Vehicles. Volume 1: Noise Sources, Acoustical Society of America through the American Institute of Physics, Woodbury. Lighthill, M.J. (1952) On sound generated aerodynamically. Part I. General theory. Proc. R. Soc., A211, 546587. Morse, P.M. and Ingard, K.U. (1968) Theoretical Acoustics, McGraw-Hill. Parry, A.B. (1988) Theoretical prediction of counter-rotating propeller noise. PhD thesis. Department of Applied Mathematical Studies, University of Leeds. Roy, D. (1983) Doppler frequency effects due to source, medium and receiver motions of constant velocity. AIAA Paper No. 83-0702. Schlinker, R.H. and Amiet, R.K. (1981) helicopter rotor trailing edge noise. NASA CR-3470. Whiteld, C.E., Gliebe, P.R., Mani, R. and Mungur, P. (1989) High speed turboprop aeroacoustic study (single rotation). Volume 1 model development. NASA CR-182257. Whiteld, C.E., Mani, R. and Gliebe, P.R. (1990a) High speed turboprop aeroacoustic study (counterrotation). Volume 1 model development. NASA CR-185241. Whiteld, C.E., Mani, R. and Gliebe, P.R. (1990b) High speed turboprop aeroacoustic study (counterrotation). Volume 2 computer programs. NASA CR-185242. Zorumski, W.E. and Weir, D.S. (eds) (1986) Aircraft noise prediction program theoretical manualpropeller aerodynamics and noise. NASA TM-83199, Pt. 3.

You might also like