Elements of compressor noise control

C. F. Speich

EI6ments de la lutte contre le bruit du compresseur Le bruit accompagne toujours le conditionnement d'air et le refroidissement, mais il ne constitue pas toujours un probl#me. Cependant 8 notre #poque de r#glementation croissante de la part des services gouvernementaux ~ tous les niveaux et d'une p/us grande prise de conscience du bruit par le client final, le bruit produit par les installations et leurs composants doit faire I'objet des pr#occupations du concepteur.

Ce rapport traite du proM#me du bruit provenant des compresseurs vo/um#triques. Le probldme est

d#fini en fonction du bruit typique, de la pulsation de pression et des spectres de fr#quences produits par les compresseurs. On pr#sente /es caract#ristiques des installations pr#fabriqu#es en deux blocs et des appareils en toiture pour montrer /'influence du compresseur sur le bruit produit par ces installations.

On examine, en fonction du compresseur, les sources de bruit, de pulsation de gazet de vibration, en examinant les caract#ristiques de/a conception provoquant et d#terminant les spectres et leur ampfitude. On indique enfin les m#thodes courantes, et route autre fa¢on d'att#nuadon du bruit, de la pulsation de gazet de la vibration des compresseurs.

Sound is an ever present by-product of air conditioning and refrigeration, but is not always a problem. However, in this age of increasing regulation by government agencies at all levels and a greater awareness of noise by the ultimate consumer, the noise generated by systems and their components should be a concern of the designer.

This paper will deal with the problem of noise from positive displacement compressors. The problem will be defined in terms of the typical sound, pressure pulsation and vibration spectra generated by compressors.

Data for split systems and rooftop units are presented to show the effect of the compressor on the noise generated by these systems.

In terms of the compressor, the sources of sound, gas pulsation and vibration will be discussed in light of those features of the design which cause and shape the spectra and determine their amplitude. Finally, common methods to attenuate or otherwise reduce the sound, gas pulsation and vibration of compressors will be outlined.

Cost, performance, reliability and noise control have always been concerns of the compressor designer. Competitive pressures have molded decisions which might affect their relative ranking but all these factors were considered. Even though there has been little change at the beginning of the 1980's, noise control has taken on a new emphasis. Regulating actions by the Federal and local governments in the US are now beginning to demand quieter products from the designer.

A compressor does not operate alone, but is a component of a larger system. Noise control is normally associated with the total system. However, since the compressor is a significant generator of sound, its effects must be dealt with both

The author is the Staff Engineer, Advanced Development Engineering, The Trane Company, La Crosse, Wisconsin, USA. This is an edited version of a paper presented at the ASHRAE semi-annual meeting in Chicago in January 1981 and is reprinted with permission. Paper received February 1981.

individually and in conjunction with the rest of the system.

The purpose of this paper is to discuss various considerations of noise control in refrigerant, positive displacement compressors. Although the specific examples given will relate to compressors in the light commerical and industrial size range, the material will be also applicable to smaller sizes as well. Like any comprehensive engineering problem, noise control in compressors is a complex study. What remedial actions are finally taken will depend on the designer's choices as to where the greatest needs are in terms of noise control.

The approach used in this paper is to first define the problem in terms of the effect of the compressor on system noise. Then the sources of noise are defined and finally some solutions to noise problems are outlined. The actual solutions may be one or more of these approaches together with a few innovative ideas.

01 40-7007/81/050281-07S2.00 Volume 4 Number 5 September 1981 © 1981 IPC Business Press Ltd and IIR 2.81

70

6O m "¢1

50

Z

I -

30

20 I I I 1 1 1 IOO 200 500 IOOO 2000 5000 I O O O O

Frequency, Hz

Fig. 1 Typical one-third octave band sound pressure spectrum of airborne noise from an industrial type compressor at 1 5 ft (4.57 m)

Fig. 1 Spectre de pression acousdque dans une bande d'un tiers d'octave du bruit transmis par/ 'a i r provenant d'un compresseur industriel 8 4.57 m

Def in ing the problem

In dealing with compressor noise, it must be first recognized that compressors emit noise in various forms. The most obvious of these is direct, air-borne sound. However, as will be pointed out later, the end result of the pressure pulsations in the discharge line as well as the vibratory motion of the compressor can also lead to noise generation within the unit. Thus, it is important to understand how each of these spectra are generated and how their effects can be alleviated.

Fig. 1 is a typical sound pressure spectrum for a mid-size, industrial type, hermetic, 105 kW (30 ton) compressor measured in a reverberant environment. It is evident that a broad band of frequencies are represented in the spectrum. The peak amplitudes are generally in the 800 to 2000 Hz frequency range. The frequencies where the actual peak amplitudes occur will be a function of the size, shape and construction of the compressor itself. However, it can be generally said that the frequencies where these peak amplitudes occur will be higher for smaller compressors.

Actually, the 1/3 octave band format used for this figure does not clearly denote the detailed character of the air-borne sound. Compressor noise is a mixture of both discrete and broadband random noise. In fact, it is this type of frequency signature content that makes the noise distinctive.

Fig. 2 is a spectrum for the discharge gas pulsation measured in the discharge line downstream of the compressor. Generally, the largest amplitudes will be at the lower frequencies. Peak-to-peak amplitudes for the waveforms associated with these frequencies will generally be in the 35 to 70 kPa (5 to 10 psi) range. Considering the density of the discharge gas,

this can represent a significant amount of energy. This energy, transmitted through the discharge piping of a unit, can lead to noise problems elsewhere in the unit.

The forces and accelerations within the compressor lead to motions of the compressor and thus to considerations of vibration. In dealing with vibrations, engineers are usually concerned with displacements or velocities at the attachment points. In either case, the peak amplitudes are generally or a tow frequency nature, often associated with the running speed and the first two or three harmonics.

Fig. 3 is a sound pressure spectrum for a mid-range commerical split system. As such, it consists of a compressor, condensing coil and condensing fans.

120

I10

~, I00

.g

9o

I--

8 0 -

70

Fig. 2

I I I I I I IOO 200 500 IOOO 2000 5000 IOOOO

Fr~l~ncy, Hz

One-third octave band pressure spectrum of the discharge gas dynamic pressure from an industrial type compressor

Fig. 2 Spectre de pression dans une bande d'un tiers d'octave de la pression dynamique du gaz au refou/ement pour un compresseur ibdustrie/

90

8O

!

~ 70 ̧ - 2

" 0 60 g,

50

Compressor and rGn

40 I I I IOO IOOO IQO00

F r e q u e n c y , Hz

Fig. 3 Octave band sound pressure levels as a function of frequency for a typical 30-ton commercial split system

Fig. 3 Niveaux de pression acoustique dans une bande d'un octave en foncdon de la fr4quence pour une installation commerciale type pr4fabriqude en deux blocs de 105 kW

282 Revue Internationale du Froid

In this case, the compressor is in a separate compartment with panels on all sides. There are three apparent sound sources in the unit: the condensing fans, the air-borne sound from the compressor modified by the effect of the compartment, and the radiation of sound from the condensing coil. It is not possible to separate the effects of the last two sources but both are attributable to the compressor since the energy radiating from the coil comes from the discharge gas pulsations. The effect of the compressor on the overall system noise is also apparent because the spectrum for the compressor plus fan is greater than that for the fan only at frequencies less than 2000 Hz.

Again, system size and configuration can play an important role in the contributions of various components to the noise. The design choices among compressor size and type, condenser coil surface area and fan selection can cause variations in the apparent sources and amplitudes of the overall system noise.

Another type of unitary product whose noise is affected by the compFessor is the rooftop unit. A common application of this product is to set the unit on a curb surrounding a hole in the roof structure. At times, the only barrier between the bottom of the unit and the listener is the space between the roof and the ceiling and the ceiling panels themselves. Tests have shown that the compressor noise heard in the occupied space is not the air-borne sound from the compressor but rather is due to the effects of the discharge gas pulsation. The energy in the pulsation is transmitted to the frame of the unit through any clamps on the discharge line and through the condensing Coil which is hard-mounted to the unit frame. This energy causes the base pan of the unit to vibrate much like a sounding board, and thus the pan becomes an apparent source of the compressor noise fr'om the Unit to the occupants below. The spectrum of this noise wil l be determined by the principal frequencies of the source plus the influence of various resonance frequencies within the structure. There wil l also be noise from the evaporator fan and its spectrum will be determined by such factors as geometry, speed, static pressure, f low rate and the influence of the duct geometry. To the occupant, the compressor noise wil l be distinctive because of its pure tones and peaked spectrum which tends to make it stand out over the more random, broad-band noise of the fan. Often, it is the character of the compressor noise that may cause customer concern and require remedial action.

Def in ing the sources of noise

When dealing with compressor noise, it is obvious that the compression process and the events associated with it are the ultimate source of the noise. Thus, we need to understand what it is about the compression process which can cause noise problems.

2 o

I I

I I

f Time

O '

Fig. 4 Sketch of the volume flow rate of a single-discharge gas pulse as a function of time

Fig. 4 Schdma en fonction du temps, du ddbit volume d'une pulsion de gaz au refoulement

In a positive displacement compressor, successive batches of refrigerant vapour f low into the compressing volume, are compressed to a higher pressure and discharged into a confined space. The f low is not steady but pulsating. Fig. 4 shows a sketch of the f low of vapour from a single discharge port as a function of time. The pulse of vapour lasts for about one-sixth of the time required for one complete rotation of the crankshaft. A harmonic analysis of this pulse would show it to be rich in the higher harmonics of the running speed. The use of valves further affects this pulse by modulating the flow. Any flutter in the valve will also affect the spectrum of the f low from the discharge port. The same type of action occurs on the suction side as well. The pulse has a different shape and lasts 2 to 2.5 times as long depending on the operating point of the compressor. However, it likewise is not steady and leads to higher harmonics of the gas flow.

Since the vapour must f low through manifolds to bring it to the compression volume and out of the compressor, the complex impedance of these manifolds is important. Because of the intrinsic geometry of the manifolds, the f low of the gas will be partially restricted in both a static and dynamic sense. In other words, the manifolds wil l be more restrictive at some frequencies than at others. These restrictions lead to pressure drops which, in turn, lead to dynamic forces acting on the surfaces which contain them.

There are other sources of excitation in the compressor as well. Fully hermetic compressors have the electric motor drive as an integral part of the compressor assembly. The fact that the compression process is pulsating and that a finite number of cylinders or compression volumes are involved, results in the motor having a varying torque requirement. As a result, the motor speed is not constant 1 during one crank rotation and there are varying forces from the stator acting on the compressor structure. These forces wil l have spectra which contain many harmonics of the running speed.

In addition, the motors themselves may be a source of noise. Webb 2 discusses how rotor design can affect the noise produced. He indicates that

Volume 4 Num(~ro 5 Septembre 1 981 283

enclosed rotor bar design, a design concept often found in reciprocating compressor motors, may produce significant levels of low frequency noise at line frequency and its harmonics up to 400 to 500 Hz. In the same reference, Fehr and Muster 3 give the means to calculate the expected frequencies for the electromagnetic forces found in the rotor and stator of the motor. How significant any of these frequencies wil l be depends on the structural response of the stator and motor to these periodic forces.

Aerodynamic noise from fan blades or non-uni form features of the rotor may also be a factor in the generation of noise. This would be expected to be broad-band and thus not a serious problem. However, rotor holes or slots have been known 3 to cause 'siren' effects at discrete frequencies in the 300 to 600 Hz range.

Mechanical noises can be a third source of noise. Here the concern is wi th the effects of the impact of valves on seats and stops and bearings. In piston- type compressors, it may also be piston slap 4 which occurs everytime the piston passes through top or bottom dead centre. In rotary-type compressors, it may be the interaction of the rotor and vanes or other components moving in the cylinder by the action of the dynamic pressures acting on the rotor. This has been discussed in detail by Johnson. 5

The clicking noise associated with the impact of valves is a common noise known to anyone who has listened to a compressor. It generally contains high frequencies due to the short times associated with the impact forces. Further, since valves may not seat exactly flat upon impact, short bursts of valve 'r inging' can occur at their natural frequencies. On occasion, it is possible to see discrete frequencies in the sound spectrum associated with natural modes of vibration of discharge valves or stops.

Webb 2 briefly discusses bearing noise in small reciprocating compressors. Such noise has been attributed to excessive journal-bearing clearances, poor surface finish and improper lubrication. Again, these phenomena can provide force excitations to the supporting compressor structure.

S t r u c t u r a l response

There are many structures wi th in a typical compressor which can respond to the forces outl ined above. The general effect is that, because of the specific natural frequencies of the various components, different frequencies of the forcing functions wil l be enhanced or diminished. As already stated, the forcing functions wilt be harmonic with the running speed. It is not uncommon to find peaks through the 1 50th harmonic. Generally, the strongest peaks wil l be at the lower frequencies with the amplitudes having lower values for higher frequencies. However, there may be a few relatively strong peaks at higher frequencies. In the final analysis, however, it is the

amplitude of the driving force coupled with the propensity of the various structures to respond which determines how noisy a compressor wil l be.

The transmission of noise depends on the presence of a source, transmission path and a receiver, tn a compressor, there are a number of potential sources as has already been outlined. The transmission path wil l be the structure which makes up the compressor and/or a refrigerant gas path. The receiver is, obviously, the person hearing the compressor.

Noise control is accomplished by affecting either the energy sources and/or the responses of the structure and gas paths. To control responses, we need to know about the energy sources and the design parameters that control the response or the structure to these sources. As an example, consider the inlet and discharge manifolds which have already been discussed. Recently, there has been extensive work on modell ing manifolds to establish the variations in impedance with frequency. Singh and Soedel discussed two general approaches which are predicated using acoustical techniques; namely, a lumped parameter approach 6 and a distributed parameter approach. 7 In the former, the manifold is assumed to be composed of a series of volumes and restrictions which are analogous to 'springs' and "masses', respectively, in a vibrating mechanical system. Classical techniques are then used to establish the impedance of such models as a function of frequency. In the distributed parameter approach, the model is divided into a series of element-type structures and matrix procedures are used to couple the elements.

Both of these techniques were developed to be used in conjunction wi th computer simulation models for compressors. They permit the models to be more comprehensive since the pressure effects of the manifolds could be introduced into the simulation. Of course, the manifold models can be used alone if only their impedance is desired. To predict pressures, the details of the f low pulse must be known. That is the purpose of the comprehensive compressor simulation - to supply an estimated pulse. The problem is that it is diff icult to obtain accurate amplitude and, phase information about the pulse. The result is that this approach often predicts frequencies reasonably well but tends to be poorer in predicting wave forms and pressure amplitudes. A better approach with the current state-of-the-art, when pressures are required, is to measure them experimentally. The models can then be used to help interpret the reasons for the predominance of specific frequencies in the pressure f ie ld

Cylinder heads are structures which can see large dynamic pressure amplitudes. Measured peak to peak amplitudes under typical heads may reach 140 to 170 kPa (20 to 25 psi). Some of the energy wil l be transmitted through the walls of the head and be radiated as sound from the outer surface. In addition, some designs of the head may have diaphragm-type vibrational modes in the frequency

284 International Journal of Refrigeration

range of 1000 to 2000 Hz. In this case, the transmission of the dynamic pressure is greatly enhanced. In accessible hermetic compressors, the transmitted energy becomes part of the air-borne sound from the compressor. In steel-shelled compressors, it enters the gas space between the compressor subassembly and the shell.

The gas space in a steel-shell compressor is also subject to dynamic pressures from the suction manifolds for the cylinders or compression volumes. The pulsating gas f lows produce dynamic pressures which communicate wi th the shell cavity. Johnson 8 has shown that it is possible to excite modes of the cavity between the compressor sub-assembly and the inside of the shell. The pure tones which result can be very loud and often occur in the 200 to 500 Hz range. The amplitudes are very sensitive to gas temperatures and pressure which affect the sonic velocity of the gas. Thus, a common symptom of the noise is the fading in and out of the tones as the operating point of the compressor changes.

The final element in the transmission path of a steel- shell compressor is the shell itself. It is a good sound barrier. The author is aware of some research work which shows the shell generally to be worth some 20 dB in terms of insertion loss at many frequencies. However, to be sure. the shell does not do a good job in suppressing noise at the various resonance frequencies of the shell. There are a number of good sources of information on shell theory as it relates to dynamic behaviour. Ingalls 9 has summarized some of this work and concluded that reasonable agreement can be obtained with experimental data if freely-supported, thin. shal low cylinders are assumed. Using such approaches, it is possible to predict mode shapes and their associated resonance frequencies as well as the effects of internal pressure and gas density. Faulkner 1° has also shown the effect of shell geometry on the resonance frequencies of thin shells.

Fig. 5 shows how the natural frequencies vary for a uniform cylinder supported at the ends wi th no axial constraint. The vibration of the cylinder at the lowest mode number, is often called the 'breathing' mode since it resembles the motion of the chest during breathing. The next frequencies occur wi th modes which divide the cylinder into several vibrating parts separated by nodal lines. The effect of these natural frequencies has also been demonstrated by Ingalls ~1 who has shown measurements in the free sound field around the compressor. Since the frequencies associated with these modes are often in the range of 1000 to 1500 Hz, they can play a very important role in the noise radiated from the compressor.

In accessible hermetic compressors, vibrational modes can be found in the housing structure which are similar to the modes described for cylinders. The author has detected the presence of such modes in the motor barrel of a large, industrial type compressor. Since motors are generally assembled with a slip fit in the motor barrel, much of the

C o n s t o n t oxiol mode

=-

I I I I I I 1 I

I n c r e o s i n g m o d e n u m b e r

Fig. 5 Natural frequency as a function of mode number for a uniform cylinder

Fig. 5 Frdquence naturelle en fonction du nombre modal pour un cylindre uniforme

structure is free to move. This type of vibration is expected to have been excited by a harmonic of the torque reactions of the motor.

Defining the solution

From a practical standpoint, achieving dramatic reductions in noise from compressors is a diff icult problem. It is that way because of the complex structures which make up compressors and because of the many harmonic frequencies which are available for excitation. Changing the geometry of a structure often only leads to changing the frequencies at which resonance occurs. To be sure, if a cavity resonance or a vibrating terminal box cover is present, specific action can be taken to control these sources. Finally, it is a 'fact of life' that noise control often comes 'after the fact', The definit ion of noise problems does not come until the first prototype is built and, by then, it may be too late to make significant changes.

For larger compressors of the semi-hermatic or industrial type, traditional materials, such as cast iron, have been used to reduce the transmission of energy through the structure and to provide some material dampening for the structure. The outside surface area is actually a series of smaller structures wi th bolted and gasketed joints. Section thicknesses and structural shapes are determined on the basis of strength and safety consideration. Modern day finite element analysis allows the designer to also establish the effect of some design changes on the dynamic behaviour of the structure. In this way, the design of the structure can be tailored to minimize structural responses in the frequency range of interest.

For steel-shelled compressors, information has already been given from which the structural frequencies of the shell can be estimated. Again, finite element analysis can be used to make these estimates more specific to a given total shell geometry. However, knowing the frequencies does not in itself guarantee a solution to the problem. Changing shell thickness or geometry may not shift

Volume 4 Number 5 September 1981 285

i I I

0 ' I I

Resonant type

Expansion chamber type

Fig. 6 Sketches of typical resonant type and expansion type mufflers used in refrigerant compressors

Fig. 6 Schemas de silencieux ~ r#sonance .et ~ chambre de d#tente utilisds dans les compresseurs frigorifiques

6 0

)n Iou

-o 5 0

~ 4o

30 Muffler output

2O

H30 500 IOOO

Frequency, Hz Fig. 7 Insertion loss asa function of frequency for an expansion chamber type muffler external to a 30-ton industrial type compressor

Fig. 7 Pour un compresseur industriel de 105 kW, perte, en fonction de/a fr#quence, due ~ /' introduction d'un silencieux ext#rieur ~ chambre de d#tente

70 ~ Compressor output

1

30O0

the critical modes sufficiently or may have some adverse economic consequences. Circumferential ribs can affect mode shapes and the use of an internal liner tack-welded on the shell can, through dampening, affect the amplitude of the vibration at the resonance frequencies. However, dampening of this type tends to raise the sensitivity of the structure at non-resonance frequencies, the overall amplitude of the noise radiating from the shell may

not change signif icantly when resonance frequencies are shifted unless a very strong mode was present in the original noise problem.

One common solution to air-borne noise radiation is to enclose the shell in either an acoustically treated, lined compartment or in a wrap of open-celled acoustic foam and vinyl. Both act as an additional barrier to the sound and, as such, represent a logical means of achieving broad band attenuation of the noise radiating from the shell. Since such devices affect the heat transfer from the shell, the operation of any internal temperature sensing devices may be affected.

Mufflers are also commonly used in compressors and may be found on either the suction or discharge side or both. When they are close to the compressor, they act as a modification to the manifolds. When separate from the compressor, they tend to act independently of the acoustics wi th in the com- pressor. In either event, the beneficial effect of the muffler is to reduce the amplitude of the dynamic pressures in the manifold and/or lines over a controlled frequency range.

Mufflers generally fall into two general groups; namely, reactive and dissipative. Reactive mufflers change the acoustic impedance of the manifold or line and impede the f low f luctuation at the frequencies to which the muffler is tuned. Dissipative mufflers provide means of absorbing some of the pulsation energy and dissipating it in the form of friction. Mufflers in refrigerant compressors are generally of the reactive type since dissipative mufflers can become contaminated by the oil normally associated with these types of compressors.

Reactive muffler types can be further divided into two general types; namely resonance type and expansion chamber types. Fig. 6 shows sketches of these types of mufflers. A resonance type muffler wil l, generally, attenuate a rather narrow band of frequencies but to a large degree. The expansion chamber type attenuates over a broad band of frequencies. Its effect can be quite great if the ratio of chamber cross-sectional area to inlet line area is large. Both types are found in commercially available compressors.

The reactive type mufflers have been designed using classical acoustical filter theory for some time. The classic work done by Davis, et al, 12 provided a very comprehensive approach to the design of resonant, expansion chamber and combined mufflers. Al though their approach did not take into account f low considerations, its use has led to some very effective mufflers. With the advent of digital computers, more sophisticated approaches have b e e n used . 13,14.15 The effect of the increased capability is to design the muffler to the specific needs of the compressor.

Fig. 7 provides an example of the effect of inserting an expansion chamber type muffler in the discharge

286 Revue Internationale du Froid

line external to an industrial type, semi-hermetic compressor. Shown are 1/3 octave band spectra of the gas pulsation before and after inserting the muffler in the discharge line. This particular muffler was designed to have its peak attenuation around 300 Hz. The differences between the two spectra is the insertion loss (attenuation). The effective attenuation occurs between 100 and 1000 Hz. If direct measurements are made of the dynamic pressure in the discharge line before the muffler, differences wil l be noted in terms of amplitude and frequency as compared to a similar measurement without a muffler in place. These differences are due to the back pressure generated by the reactance of the muffler.

This effect of back pressure is quite important especially if the muffler is close to the discharge valves as it would be if the muffler were an integral part of the cylinder head. The effect of this back pressure can be such as to adversely affect the f low of gases through the valves. Thus mufflers can, in some cases, cause performance losses in the compressor due to this back pressure effect. These losses are in addition to any f low losses which might occur due to fluid dynamic effects within the muffler itself. The designer must be careful as to how and where the muffler is applied in order to minimize such losses.

Vibration control is important as it relates to the attachment points on a compressor such as at the feet. Large grommet-type structures or ribbed elastomeric materials are commonly used to resiliently mount the compressor relative to its base. There are a number of manufacturers' guidelines which can be used to select such isolators as well as the ASHRAE Systems Handbook. 16 It is important to remember that both the vertical and horizontal stiffnesses are important. The natural frequencies of the compressor on the isolation mounts should be calculated. These frequencies should be lower than the running speed of the compressor by at least a factor of three. Since this implies low stiffnesses, there may be a requirement to constrain the isolators during periods of shipment and/or start-up and shutdown.

Conclusion

Noise control in positive displacement compressors is not an easy task. It involves the forced response of a complex structure. However, the use of traditional materials and structures as well as

mufflers and enclosures do permit practical means of controlling noise. In addition, new computer oriented techniques permit the engineer to design the compressor with noise control ~ in mind. Combining the old with the new should lead to quieter compressors and, therefore, quieter systems in the future.

References- 1 Jensen, E. H. Effect of compressor characteristics on

motor performance, ASHRAE Transactions 66 Paper 1709 (1960) 194-201

2 Webb, H. E. Compressor. household-refrigerator, and room air conditioner noise. Handbook of Noise Control, McGraw-Hill, NY (1957) Chapter 28, 28-3

3 Fehr, R. O., Muster, D. F. Electric-motor and generator noise. Handbook of Noise Control. McGraw-Hill. NY (1957) Chapter 30. 30-4 and 30-12

4 Ungar, E. E., Ross, D. Vibrations and noise due to piston-slap in reciprocating machinery, Journal of Sound and Vibration 3 2 (1965) 1 32-146

5 Johnson, C. N., Hamil ton, J. F. Fractional horsepower, rotary-vane, refrigerant compressor noise study, Proceedings of the 1972 Purdue Compressor Technology Conference. Purdue University (1972) 74-78

6 Singh, R., Soedel, W. Interpretation of gas oscillations in multicylinder fluid machinery manifolds by using lumped parameter descriptions, Journal of Sound and Vibration, 64 3 (1979) 387-402

7 Singh, R., Soedel, W. Mathematical modeling of multicylinder compressor discharge system interactions, Journal of Sound and Vibration 63 1 (1979) 125-143

8 Johnson, C. N., Hamil ton, J. F. Cavity resonance in fractional H P refrigerant compressors Proceedings of the 1972 Purdue Compressor Technology Conference, Purdue University, (1972) 83-89

9 Ingalls, D. J. Understanding noise radiation from a hermetic compressor, Proceedings of the 1972 Purdue Compressor Technology Conference, Purdue University (1972) 69-73

10 Faulkner, L. L. Hermetic refrigeration compressor acoustics, Purdue Noise Control Conference, Purdue University (1971 ) 107-114

11 Ingalls, D. J. Curtailing hermetic reciprocating compressor noise, ASHRAE Journal 13 (April 1971 ) 42-47

1 2 Davis, D. D. Jr, Stokes, G. M., Moore, D., Stevens, G. L. Jr Theoretical and experimental investigations of mufflers with comments on engine exhaust muffler designs, Report 1192. National Advisory Committee for Aeronautics, 1954. See also: Davis, D. D. Jr Acoustic filters and mufflers, Handbook of Noise Control, Mc-Graw- Hill, NY. Chapter 21

13 Hundley, T. C. Refrigerant muffler analysis Proceedings of the 1976 Purdue Compressor Technology Conference, Purdue University (1976) 382-388

14 Mi l ler, D. F., Holten, B. W. Muffler analysis by digital computer, ASHRAE Transactions 66 Paper No. 1710 (1960) 202-216

15 Alfredson, R. J., Davies, P. O. A. L. Performance of exhaust silencer components, Journal of Sound and Vibrations 15 2 (1971) 175-196

16 ASHRAE Handbook. 1980 Systems Volume. Sound and Vibration Control. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (1980) Chapter 35, 35.25-35.27

Volume 4 Num6ro 5 Septembre 1981 287