Purdue University Purdue e-Pubs
International Compressor Engineering Conference School of Mechanical Engineering
1974 Sound Power Measurements on Large Compressors Installed Indoors - Two Surface Method G. M. Diehl Ingersoll-Rand Company
Follow this and additional works at: https://docs.lib.purdue.edu/icec
Diehl, G. M., "Sound Power Measurements on Large Compressors Installed Indoors - Two Surface Method" (1974). International Compressor Engineering Conference. Paper 126. https://docs.lib.purdue.edu/icec/126
This document has been made available through Purdue e-Pubs, a service of the Purdue University Libraries. Please contact [email protected] for additional information. Complete proceedings may be acquired in print and on CD-ROM directly from the Ray W. Herrick Laboratories at https://engineering.purdue.edu/ Herrick/Events/orderlit.html SOUND POWER MEASUREMENTS ON LARGE COMPRESSORS INSTALLED INDOORS - TWO-SURFACE METHOD -
George M. Diehl Manager, Sound and Vibration Section Ingersoll-Rand Research, Inc. Phillipsburg, New Jersey
Sound power ratings of machinery are be 3. Sound power levels coming important for a number of large machines of reasons. are difficult to obtain in most industrial Industrial installations must comply with locations. recently enacted state and local noise con trol codes, and sound power ratings are This situation is changing, needed to predict compliance. and sound In addi power ratings of machinery are tion to this, the Federal Noise becoming Control more important. Recognizing this, the Act of 1972 requires labeling of machinery International to show maximum Organization for Stand noise emission, and this, ardization has decided that logically, may be in terms all machinery of sound power. noise measurement standards must include The a procedure for determination of sound sound produced by a compressor cannot power level. be stated in terms of A-weighted sound level alone. Since sound level decreases The objective, then, is to with distance from the source, develop a this dis procedure for determining the sound power tance must be given, along with the height level above of a machine when it is operating the floor where sound measurements in an industrial were taken. Furthermore, environment. This can measured noise be done by the two-surface method. depends upon the environment where the compressor is installed. A product noise For accurate results, sound emission label in terms of A-weighted power level should be measured in either a free field, sound level would have to show all this an anechoic additional information. chamber, or a reverberant field. Most large compressors, and simi lar industrial In contrast, the sound machinery cannot be sound power level of a tested in any of these three ~chine is independent of measurement environments. distance They must be measured where they are from the source, and although installed, radiated sound power on a customer's property, or depends on the in some cases, on the acoustic impedance into which machinery manu it is ra facturer's test stand, prior to shipment. diated, it is practically independent of environment for most machinery installa Unfortunately, neither the customer's tions. These properties are the ones that plant make sound power nor the manufacturer's test so desirable in expressing facility is the noise rating of a machine. ideal for sound measurements. In fact, some machines are installed in locations where it is simply In spite of this, sound power level is impossible not to conduct a sound test that has any often stated, nor even measured for meaning. large machinery. Some machines are nearly as The air conditioning large as the room in which industry and certain gas turbine they are lo manu cated. Often other nearby equipment, facturers provide sound power information, which but almost cannot be shut down makes more noise all other large machinery is than the machine rated in terms of near-field being tested. In other sound pressure instances, it is too dangerous to levels. There are several reasons for try to this: take sound measurements over the top of large rotating machinery, as required in sound power tests. 1. The-Occupational Safety and Health Act states hearing damage criteria in Not all cases are imPossible. There terms of A-weighted sound level. Most are many machines that can be sound-tested on machinery sound specifications require the job this information. and both octave-band sound pressure levels and octave-band sound power levels 2. can be obtained with reasonable Until now industrial plant operators accuracy. have had Acousticians and noise control little or no need for sound engineers cannot power ratings. back away from these projects because they cannot be done with
230 laboratory accuracy. Something must be room done, and they must do it with the best accuracy possible under actual operating SPL - PWL 10 log~r2 + .!}+ 0.1 conditions. Measurements in these semi reverberant environments are of great where, practical importance to industry. SPL is sound pressure level, in decibels, 2 Sound power levels must be calculated re 2.0 X 10-SN/m : from measured sound pressure levels. Since sound pressure levels are influenced PWL is sound power level, in decibels, by the environment in which the measure ments are made, it becomes necessary to re lo-12w; know the effective room constant of the environment. There are several ways to r is the distance from the source, in determine this: meters; and, R is the room constant, in square meters. First, it may be calculated by estimating the average absorption coefficient of This equation can be plotted in a series the floor, ceiling, side walls, and of curves for various combinations of other items in the room, and using the r and R. appropriate equation for room constant. Although this technique can yield reason able accuracy for some relatively simple areas, it turns out in the case of most industrial plants that it is no better than simply looking at the location and est imating the room constant directly. In other words, it is not very accurate. Second, it may be determined by measuring the reverberation time with a microphone, sound level meter, and graphic level re- · corder. This method is usually not feasible in industrial area. It involves the use of a noise source that is stopped quickly; the time for the level to de crease by 60 dB is measured. High ><•DISTANCE FROM ACOUSTIC CENTER machinery noise precludes the use of most sound sources; attempts to get around it, such as by firing a gun as the noise It can be seen from these curves that in source, are usually not acceptable to the case of ·a small source, the decrease plant operators. in sound pressure level with distance provides a means for determining the Third, the effect of the environment may effective room constant of the area. be determined by either a calibrated sound source and an absolute comparison With this reasoning, a set of correction test, or by an auxiliary sound source in factor curves can be obtained by plotting SPL - SPL on the vertical axis and r, a relative comparison test. In these 1 r methods, the machine under test should be the distance from the source, on the moved out while measurements are being horizontal axis. SPL is the octave made on the reference source. This is band sound pressure level1 at a distance obviously impossible, and the technique of 1 m from the nearest major surface of loses its value when it is found that the machine, and SP.Lris the octave-band many different answers are obtained by sound pressure level at a distance of placing the reference sources at various r (in meters). locations with respect to the machine under test. Furthermore, the sound pro The curves, then, can be used to obtain duced by availabe reference sources is a correction factor to be subtracted much too low to make them usable in most from the octave-band sound pressure industrial areas. levels measured in the semi-reverberant room. From this, the approximate level There is one technique that is feasible that would be measured in a free field and that is to use the machine itself as can be obtained. the sound source. To check the correction factor's validity, First consider the equation relating an ILG sound source was tested in a free sound pressure level, sound power level, field. Next, it was tested indoors in and room constant in a semi-reverberant
231 10 !--""' . - -·---.... / -~ <6/ / V/ ,,o ~ c- //v .. I-"" L/'v - ? ..0 .1 a.. v-- I ~ "'.1 V/ a.. ~ "' 25 / l-- -~ -- f-'""" ~ ,_v- 4.0 ~-5 V::= ~ 5.0 ...... 5.5 t- 1(WJ 1000 1.. 0. I-- FRE:QUE::NCY v::: 1--~0 IN CVCL5$ PER SECONO 1 v::= QCT,i,VE ,AM MND~ IN CYCLES ron !!~ND 2 J 4 s 86 9 7 10 110 - ~JJ - 11" - !
232 instead of from the nearest surface of the Then, 2 machine. Application. of the second cor 2Tfr =TTa(b + c) rection factor did net- seem to be just ified, however; actually, it seemed to r.take the compa-rison worse in most cases. 2 = a(b + c) r 2 A criterion frequently used in sound power level determinations requires that, in field, order to be out of the near r"" ~(b ; ~\ measurements should be made a·t a distance not less than twice the largest dimension test. In measure 3. calculate the average sound pressure of the machine under each octave band of over a reflecting plane, the level, NPL, in ments made interest; and, minimum distance is sometimes increased times the largest dimension of to four 4·. calculate· the sound power level, the source. assuming that measurements had been made from the acoustic may be followed in labo at a distance of r .This criterion using ratory tests on small machines, but it center, never can be met in industrial 12 almost 20 log r + 7.80, re lo- w. plants. Furthermore, recent investigations PWL =· SPL + have shown that. reasonable accuracy can distances as There are several deficiencies with this be obtained with-microphone there is no correct ioit small as 1/4 m from the machine. method'-. First, for the environment, and caLculated be more One method that has been proposed flor de sound power levels can never sound power level of a large accurate than the sound pressure level-s termining the · machine, using near-field measurements, from- which they were calculated. is the follmiTing: Another deficiency is·that in the case of dimensions, the 1. measure sound pressure levels· around· machines with certain the top at a distance distance r turns. out to. be less than the the machine and over sound from the nearest major surface of machine dimension, meaning that of 1 m calculated on the the machine, assuming that the "leasure power levels were on- the surface of. one· half basis of an average sound pressure ments were made machine. of an elliptical cylinder over the level·, measured inside the is physically impossible. machine~ This, of course.,. The two-surface method for determ·ining sound power level overcomes these de SOUND POWER MEASUREMENT ficiencies .. It provides an adjustment· LARGE MACHINERY for room environment: it does not rely on True, it cannot AREA OF EQUIVALENT HEMISPHERE an "equivalent radius". be applied in all cases, but when it can be used·, it is- as- accurate, or more accurate, than other industrial sound power procedures. In this method, octave-band sound pressure levels are measured on the surfaces of two hypothetical parallelepipeds over the machine being tested. The second imaginary "box" is farther away from the machine than the first, and therefore , is greater than that of the its area, s 2 first, sl.
Although the sound power level is constant, 2. calculate the radius of a hemisphere the average of the sound pressure level that has the same surface area as the half measurements for area s will be greater cylinder- 1 area s because s lateral area of entire elliptical- cylinder than the average for 2 1 of end X length - = perimeter is closer to the source than area s 2 perimeter of elliptical end =Tt(b + c) lateral area of cylinder = 2JTa(b + c) lateral area of half cylinder "=TTa(b +c) area of equivalent hemisphere = 2JTr2
233 ·------·d I ---_____ !______I I to use a great many measurement points. 1 To demonstrate I I this, several different compressor types with entirely different operating principles were tested in three : i ways to determine their I I rl t-4 octave-band l·i-:I I sound power levels.
First, a six-point hemispherical array was used. Second, microphone locations were arranged d in a 17-point half-cylinder ------array. Finally, 1 a I microphones were located I ,------, I as stated in ANSI S 5.1, "Test Code For The Measurement I I I 1 Of Sound From Pneumatic I Eguipnent". This code prescribes I I I I micro I ..,1., phone locations at each end and at the I I I I centers of the sides of the machine, plus I '------' I a point at the location of maximum dBA. ~------~-~----~ The dBA readings were identical in all three methods. The differences in oct ave-band sound power levels between any two methods were usually 1 or 2 dB.
FREE FIELD SOUND POWER LEVEL RE 10 ~ 12 W where, MACHINE AV25 SPL is the average of the sound pressure 1 CD CD CD 0 0 0 level measurer11ents for area s : Octava 6 Point 17 Point Difference AlliS I Difference 1 Band Homlop~ere Hall-Cylinder CD-0 55.1 0-0 SPL is the average of dBA 108 108 2 the sound pressure 0 108 0 63 90 89 -1 90 0 125 level measurements for area s ; and, 87 89 +2 87 0 2 250 106 106 0 105 -.1 500 R is the room constant. 106 107 +1 107 +1 1K 100 101 +1 100 0 R can be eliminated from 2K 97 these two equa 98 +1 97 0 tions, and sound power •K 99 96 -3 9. -5 level can be shown BK to egual 86 85 -1 85 -1
PWL = SPL + 10 log - C 1 s1 Conclusions can be drawn: 1. Tests are underway to Where, accumulate- data in the hlo-surface method to prove its accuracy. All indications are that this C = 10 log K - sl]. new procedure ~ 1 ~ s I will be the best way to de 2 termine sound power levels of large ma chinery installed indoors ~n~, itJ actual in 10 (SPL1 -SPL 2 )/10 dustrial and semi-reverberant locations. 2. Measurement of sound pressure The environmental levels correction, C, may be on two separate hypothetical surfaces pro calculated mathematically, or it may be vides a means for evaluating the en found nore conveniently from a set of vironmental correction. curves relating the correction factor 3. Previous tests have shown that the "dB to SPL - SPL and the area ratio ;s drop-off" can give 1 2 s 1 2 • a fairly reliable measure of the effect of the environment. 0 The two-surface .' method utilizes this c-...,_ principle • .' 4. '-o " Except for cases where strong direct !II ' ' --1-5 r-... ~""""- ivity effects are encountered, a large ~~ ~ r-::::-- 1'0\ number of microphone locations are not I ' 3:0 1--. ~' required. /( ' ~5 ".'l\' s. It is ' --- understood'that in conducting =4f ~--...::: t':"" ~~ these tests, the usual precautions and ' good engineering r-::::::----.-.... practice must be observed, :J ~ ::s~ such as separation or isolation of un wanted noise sources, provision of ad ditional sound absorption on hard Accuracy re of the method can be improv~d·, flecting surfaces, and ensurance that theoretically, by increasing the number of background noise does not interfere. measurement locations. 1 In pract:i,ce, it turns out that it is generally unnecessary
234