Aircraft noise is hardly a new subject as evidenced by the following note received by a predecessor of United Airlines in about 1927.
Although internal noise was the major preoccupation of aircraft acoustic engineers for many years and still is important, the noise produced by the aircraft engine and experienced on the ground has become a dominant factor in the acceptability of the airplane. With the development of high bypass ratio engines, noise due to other sources has become important as well.
Internal noise is treated by placing the engines to minimize the noise directly radiated to the cabin, (e.g. using the wing as a shield) and by providing insulating material over the entire surface of the flight and passenger compartments. If the engines are mounted on the fuselage, vibration isolation is an important feature. In the late 1980's when prop-fans were being developed, internal noise become an important consideration again. It was, at one point, estimated that 2000 lbs of additional acoustic insulation would be required to reduce cabin noise levels to those of conventional jets if prop-fans were placed on the aircraft wings. This is one reason why many prop-fan aircraft were designed as aft-mounted pusher configurations.
External noise is affected by the location of the source and observer, the engine thrust, and a number of factors that influence the overall configuration design. These will be discussed in detail later in this chapter, but first we must understand the origins of noise and its measurement.
A sound wave carries with it a certain energy in the direction of propagation. The sound becomes audible because of energy which originates at the source of the sound vibrations and which is transported by the sound waves. The changes in air pressure which reach the eardrum set it vibrating; the greater these changes, the louder is the sound.
The intensity of sound, I, is the quantity of energy transferred by a sound wave in 1 sec through an area of 1 cm. For a plane sine wave:
I = p2 / 2 r c
p = the amplitude of the varying acoustic excess pressure
r = air density
c = speed of sound
I is usually expressed in ergs per cm2 per sec. (mW/m2)
The human ear responds to a frequency range of about 10 octaves. It responds to air vibrations whose amplitude is hardly more than molecular size; it also responds without damage to sounds of intensity 1013 to 1014 times greater without damage.
The response of the ear is not proportional to the intensity, however. It is more nearly proportional to the logarithm of the intensity. If sound intensity is increased in steps of what seem to be equal increments of loudness, we find that the intensities form a sequence of the sort 1, 2, 4, 8, 16, .... or 1, 10, 100, 1000 not 1, 2, 3, 4, ... or 1, 10, 19, 28, ... . Since the ear responds differently to different frequencies, the logarithmic relation of intensity to loudness is not generally perfect, but it is easier to handle than the enormous numbers involved in the audible intensity range. Therefore, the intensity level of sound is defined in decibels as 10 times the logarithm of the ratio of the intensity of a sound, I, to a reference level defined as 10-9 erg/cm2/sec.
Thus: Sound intensity level (SPL), decibels = 10 log10 I / 10-9
The response of the ear is not exactly proportional to the decibel scale. In addition to the physical quantities, intensity and frequency, the psycho-physiological quantities of loudness and pitch must be considered. The loudness of a sound depends both on intensity level and frequency; pitch depends chiefly on frequency but to some extent on intensity. Contours of equal loudness for the average person are plotted in the following figure from Ref. 2. The actual contour values are the values of SPL at 1 kHz.
Contours of equal loudness, plotted against intensity and frequency for the average ear.
In an attempt to develop a noise measuring scale more responsive to these characteristics of the ear, the "A" scale was defined to weight noise at frequencies above 1000 Hz more heavily. Noise measured on this scale is given in units of db(A).
Frequency response weighting for the "A" scale. (From Peterson and Gross, 1967, p.9).
The scale most often used for aircraft noise measurement is the Perceived Noise Level (PNL) scale. The scale requires that the SPL be measured in each of nine contiguous frequency ranges and combined according to a special prescription, not too different from the A-weighting method, to provide a noise indication level. The units are PNdb.
The effective perceived noise level, EPNL, accounts for duration and presence of discrete frequency tones. It involves a correction factor that adds to the PNL when there are discrete tones in the noise spectrum. It also includes a correction obtained by integrating the PNL over a 10 second time interval. (Details are given in the full text of FAR Part 36.)
The effective perceived noise level correlates with people's perceived noisiness as shown in the figures below.
Subjective Reactions to Various Noise Levels
The fact that people's perception of noise varies logarithmically with sound intensity results in some interesting relations. Note that as intensity is reduced by 50% the SPL changes by 10 log I1/I2 = -3db. From the plot above this reduction would be only barely perceptible. This is why noise reduction is a challenge. To make something seem about half as noisy requires a reduction in SPL by about 10 db. This is a reduction in I of about 90%!.
People's reactions also depend on how often such noises occur and a variety of methods for averaging noisiness have been used. Sound exposure levels (SEL), noise exposure forecasts (NEF), and Day-Night-Levels all involve some kind of averaging of multiple noise events, usually with higher weightings (e.g. 10-20 times) for night flights. These are intended to capture the community response in a statistical way. (See figure below.)
Community Response to Different Noise Levels
The U.S. Environmental Protection Agency (EPA) uses a Day-Night Average A-Weighted Sound Level metric known as DNL as a method for predicting the effects on a population of the long term exposure to environmental noise. The DNL metric is legislated to be the single system for measuring aircraft noise impact and for determining land use compatibility.
Noise maps typically depict the DNL 65dB contour as this is identified by federal guidelines as the threshold level of aviation and community noise that is "significant". In general, most land uses are considered to be compatible with DNLs less than 65 dB.
Sample of Estimated Noise Footprints Atlanta Airport in Jan. 2000
Contours of constant DNL or EPNdB are often plotted to determine the areas affected at a given levels. Different aircraft may have very different footprints, this is especially obvious when comparing 2 vs. 4 engine aircraft, because of different climb rates.
Aircraft noise is generally divided into two sources: that due to the engines, and that associated with the airframe itself. As higher bypass ratio engines have become more common and aircraft have become larger, interest in airframe-related noise has grown, but engine noise still accounts for most of the aircraft external noise. The relative importance of various noise sources is shown in the figure below.
Engine noise includes that generated at the fan inlet and exit, the combustor core, the turbine, and that caused by jet mixing. While jet noise, caused by the turbulent mixing of the high speed exhaust with the ambient air, is a broad band noise source, with most of the energy directed aft of the engine at a 45 degree angle from the engine axis, the turbomachinery noise often includes discrete tones associated with blade passage frequencies and their harmonics.
Jet noise levels vary as the sixth to eighth power of the jet exhaust velocity as shown in the figure below. Early turbojet engines had exhaust velocities of nearly 2000 ft/sec and noise suppressors were used to try to obtain better mixing and lower the noise associated with the strong shear. Such suppressors were effective in reducing the low frequency noise, but often not the high frequencies and added weight and cost to the design.
The jet velocity was reduced considerably as the bypass ratio increased. This is indicated by the figure below that applies to older engines, but is still representative of the trend observed for larger modern engines.
The net result is a substantial reduction in the noise due to jet mixing. At the same time, though, the larger fan noise become more significant as seen from the figure below.
Computational aerodynamics is getting to the point of predicting such effects in a practical way, but it is a very complex problem, involving internal unsteady flows and propagation estimates.
Without such CFD tools, one can still estimate the effects of engine thrust levels, separation distances, and number of engines by scaling experimental results according to the fundamental physics of the problem as described in the following sections.
In addition to the engine noise, the shear of the boundary layer and unsteady vortex shedding from landing gear, landing gear doors, and other separated flows as well as flap edge flows contribute a significant part of the acoustic energy, especially for large aircraft on approach.
The figure on the right shows that these noise sources were still well below the requirement, but the figure was drawn in 1974. Stage 3 noise regulations now make airframe noise a significant issue.
With substantially more stringent noise regulations and a desire to reduce community environmental impact, engine companies, aircraft manufacturers, and government agencies have continued to look for ways to reduce aircraft noise.
NASA work as part of their advanced subsonic technology program includes the objective of 10 decibel (dB) community noise reduction relative to 1992 production technology. This includes:
To accomplish this, engineers are developing higher bypass ratio engines to reduce exhaust velocities, continuing to improve nacelle treatments, and operating the aircraft with take-off power cutbacks and 2-segment approaches.
The picture below shows a large acoustic test facility used by NASA Lewis as part of their work on engine noise reduction.
Noise regulations in FAR Part 36 Stage 3 include restrictions on noise in 3 conditions. The take-off noise is defined as the noise measured at a distance of 21,325 ft (6500 m) from the start of the take-off roll, directly under the airplane. The sideline noise is measured 1476 ft (450 m) from the runway centerline at a point where the noise level after liftoff is greatest. The approach noise is also measured under the airplane when it is at a distance of 6562 ft (2000 m) from the runway threshold. For each of these conditions the maximum noise level is a function of maximum takeoff gross weight, and for the take-off case the limits depend also on the number of engines. The figures below summarize the requirements.
Stage 4 noise regulations are applicable to new type designs introduced after January 1, 2006. Existing aircraft will be able to operate under Stage 3 regulations. This new Standard will be "Chapter 4" in ICAO Annex 16 and are related to the Stage 3 / Chapter 3 regulations as follows:
- A cumulative margin of 10 dB relative to Chapter 3
- A minimum sum of 2 dB at any two conditions
- No trades allowed
We start with a measurement of the noise due to a known engine at a known distance away. For example, a 25,000 lb (sea level static take-off thrust) turbofan engine with a bypass ratio of 6 produces a noise of about 101 PNdb at a distance of 1000 ft. This assumes some level of noise suppression (about 5PNdb). We might also infer a baseline engine noise from measured data such as that provided by GE and shown below:
Examples of measured noise data form reference (from GE)
We are interested in the effect of design changes on the noise, so starting from the reference value, we make corrections for thrust level, distance, ground attenuation, and noise duration. These effects are shown in the plots below and further described by an example computation that follows.
The effect of thrust level on noise is obtained by simply scaling the sound intensity (I) by the ratio of thrust to reference thrust. This correction is applied to scale the engine size or the number of engines. This means that if the engine technology is similar, reducing the installed thrust by 50% will lead to a noise reduction of about 3db. (10 log (1/2) = -3)
If thrust is reduced, not by scaling the engines, but by reducing the throttle setting, the noise is reduced much more because the fan tip speeds and exhaust velocity are reduced.
The sound intensity varies roughly as the inverse square of the distance from the source. This means that for each doubling of the distance, we expect a 6db reduction in the noise level. However, atmospheric attenuation adds about 1.2 db of reduction per 1000 ft so that increasing the distance from 1000 ft to 2000 ft results in about 7.2db attenuation. Both of these effects are included in the above plot. The presence of various obstacles and absorbing material near the ground is sometimes taken into account by adding 25% to the actual distance and considering this an effective distance.
To obtain EPNdb we typically reduce the PNdb level by about 4db for the take-off and sideline calculations and by about 5db on approach. (This reflects typical tone and duration corrections under these conditions.)
Finally we add the airframe noise which is very difficult to estimate, but which we take here to be related to the log of the aircraft weight: Airframe Noise (db) = 40 + 10 log W, where W is the aircraft weight in lbs. This fit is based on some simple scaling rules suggested by energy considerations and some empirical data from NASA and Lockheed measurements. It is very rough and applicable only on approach, but usually is not the major part of the noise contribution.
Base = 101 PNdb, 25,000 lb thrust, 1 engine, 1000ft + 4.8 for 3 engines + 1.9 for 40,000 lb SLS thrust engines - 4.0 for 1500 ft altitude at 6500m from start of take-off - 4.0 correction to EPNdb on take-off ---------- Total: 99.7 EPNdb (Flight measurement shows 98 db)
Base = 101 PNdb, 25,000 lb thrust, 1 engine, 1000ft + 4.8 for 3 engines + 1.9 for 40,000 lb SLS thrust engines - 6.5 for 1476 ft (450m) from centerline (effective distance = 1476*1.25 = 1845ft) - 4.0 correction to EPNdb on take-off ---------- Total: 97.2 EPNdb (Flight measurement shows 96 db)
Base = 101 PNdb, 25,000 lb thrust, 1 engine, 1000ft + 4.8 for 3 engines + 1.9 for 40,000 lb SLS thrust engines + 9.1 for 370 ft altitude at 6562 ft (2000m) from runway - 7.0 correction for 45% throttle - 5.0 correction to EPNdb on approach Engine subtotal: 104.8 db Airframe: 94.8 db at a landing weight of 300,000 lbs ---------- Total (add I's): 105.2 EPNdb (Flight measurement shows 106 db)