TURBOMACHINERY NOISE RATING

by
George M. Diehl
Manager, Sound & Vibration Section

Ingersoll-Rand Company

Phillipsburg, New Jersey

of enforcing them, in the way they have been written, have not
George M. Diehl, manager of the
been clearly thought out. This applies to both manufacturers
Sound and Vibration Section, and purchasers alike.
Ingersoll-Rand Research Inc., is a regis­
tered professional engineer, and a For example, one product noise specification set 80dBA as
graduate of Lafayette College, Easton, the maximum permissible noise level. When asked why 80dBA
Pennsylvania, with the degree of B.S. in was necessary, the reply confirmed that 90dBA was really what
E.E. was required. But because of possible manufacturing and mea­
He is a member of numerous commit­ surement tolerances it was decided to set 85dBA as the goal,
tees on acoustics in the International Or­ and then a 5dBA factor of safety was included, making the
ganization for Standardization, the specification level 80dBA. This could make a tremendous dif­
American National Standards Institute, ference in cost, as well as how the sound control is accom­
Compressed Air and Gas Institute, Construction Industry plished.
Manufacturers Association, and the American Society of It seems appropriate, therefore, to examine some of the
Mechanical Engineers. factors involved in setting acoustic design goals, in manufactur­
He is a member of the Acoustical Society of America, an ing quiet machinery, in testing it to confirm compliance with
Initial Member of the Institute of Noise Control Engineering, the specification, and in expressing the actual noise rating.
and the American Society of Mechanical Engineers. Noise control objectives, and economic consequences should
He has written approximately 60 to 70 technical papers be understood. Federal, State, and local noise control legisla­
and articles on sound and vibration, and is the author of a book tion, some already enacted, and others imminent, make this
"Machinery Acoustics." analysis not only desirable, but extremely important.

Relative costs and values must be assessed

ABSTRACT 1. What percentage increase in price would you be will­

ing to pay for a quiet design offering a noise reduction of, say,
An expression of the noise rating of a machine is an impor­
15dBA?
tant part of its performance specification. Precautions should
be taken, however, to make sure that the noise emission speci­ 2. Would you pay more for 15dBA reduction by internal
fication keeps the true objective in mind, and is neither too design changes than you would for 15dBA by an external acous­
stringent nor too lenient. tic enclosure? How much more?

It is a costly mistake to demand more detailed information 3. Would you be willing to accept a decrease in operating
and more noise reduction than is necessary, but the reverse is efficiency to obtain a quieter design using no enclosure or
also true. Therefore it is important to know when to require lagging?
octave band or third octave band data, instead of overall 4. \Vould you pay more for a firm sound guarantee than
A-Weighted levels, and when to request a firm sound guaran­ for simply an estimate of the noise?
tee instead of estimated noise emission levels.
5. Would you pay more for a sound guarantee in terms of
The intended use of the noise rating information should octave band levels than you would for one in terms of overall
determine whether sound pressure levels are needed, or dBA? Would you pay still more for% octave-band data?
whether extra effort is justified to obtain sound power levels.
Because of the increased emphasis on sound power ratings, the 6. Would you pay more to get sound power levels than to
get sound pressure levels?
"two-surface method" deserves consideration as the best tech­
nique to use in industrial environments. An Appendix is pro­ 7. What accuracy would you expect if you wanted to mea­
vided for· the reader who may not be familiar with the defi­ sure the noise after the machine had been installed on your
nitions of all the terms. property, to verify the guarantee?

If you were going to purchase a large centrifugal compres­ 8. Would you be willing to pay a higher price for a noise
sor or pump, would you rather have a quiet one or a noisy one? emission guarantee stating octave band levels accurate to plus
The answer to this question is obvious to both machinery man­ or minus 2 decibels than you would for accuracy of plus or
ufacturers and purchasers, and consequently a statement of minus 5 decibels?
maximum acceptable noise emission is included in practically
all machinery performance specifications. Noise reduction is costly
The answers to several other pertinent questions are not Anyone who has worked on machinery noise reduction
quite as obvious. Some sound specifications are copied, word knows that in many cases it is fairly eAsy to get several decibels
for word, from other sound specifications. Technical errors that reduction with very little effort, simply by locating and iden­
appear in one, appear in the others, and it seems in some cases tifying the major noise source and making a rather obvious
that the objectives of the specifications, and the consequences improvement.

63
64 PROCEEDINGS OF THE FIFTH TURBOMACHINERY SYMPOSIUM

The next few decibels are more difficult to obtain. More in dBA only, because errors may result if sounds are combined
looking, more identification, more recording, and narrower­ on an overall basis.
band analyses are necessary.
More instrumentation is needed, and more time is re­
After this the going gets really rough. Many noise sources quired to obtain octave band data than to read a simple
all seem to have the same sound level, and reducing one, or A-Weighted sound level.
several of them, has no effect on the overall noise, because
they combine logarithmically, and all of them must be reduced
Octave-band sound guarantees may be difficult to meet
before the overall level is improved. This is why the costs of
noise reduction increase exponentially as the acoustic goal gets Problems may arise when a purchaser demands a firm
lower and lower. guarantee that octave band levels will not exceed his specifica­
tion when the machine has been installed on his own property.
It is a mistake to demand more sound control than you He may be interested in only dBA, and has written his octave
actually require. The trick is to know what you really need. band sound specification so that the octave band levels add up
Unknown future permissible sound levels in the Occupational correctly to the overall dBA he wants; but now the guarantee
Safety and Health Standard; unknown product noise emission
becomes much more difficult to meet.
levels to be established by the Environmental Protection
Agency; and not knowing whether Federal Standards will pre­ A number of different sound spectra can combine to the
empt state and local regulations cause purchasers to anticipate same overall dBA (1), as shown on Figure 1, but guaranteeing
lower and lower permissible levels and thereby write more not to exceed a certain level in each of eight, or nine octave
stringent machinery sound specifications than they would bands means that the manufacturer has nine chances to be in
otherwise. This increases costs. violation instead of one - even though he has met the re­
quired overall dBA level.
Noise control by design not necessarily best
The room constant at the purchaser's installation may be
Everyone agrees that it is better to design quiet machin­
different than at the manufacturer's plant
ery than to rely on acoustical enclosures and lagging after the
machines have been built. Maintenance is easier, the equip­ Purchasers are interested in the noise a machine produces
ment is accessible for inspection checks and vibration mea­ in their own plant, and not on a manufacturer's test stand. But
surements, and heat dissipation problems are minimized. But the sound pressure level measured in the purchaser's plant,
when large, high �peed, high horsepower turbomachinery has depends upon the room environment there, and how rever­
been designed, developed, and tuned to maximum efficiency, berant it is. It may be different than the environment in the
machinery manufacturers, and purchasers as well, are reluc­ manufacturer's plant, and the measured sound levels there can
tant to do anything internally that may reduce that efficiency. not be guaranteed, unless the actual "room constant" is known.

Certain design changes, effecbve in reducing noise, are For this reason, manufacturers usually state estimated or
also accompanied by a sizeable reduction in efficiency. Most guaranteed sound levels "under free-field conditions." What
purchasers do not want to accept this when'they realize that a they really mean is "under conditions of a free-field over a
few points in efficiency of a large machine means greatly in­ reflecting plane."
creased operating costs for the life of the machine. Their reluc­ Free-field conditions cannot be obtained in the manufac­
tance is justified when it is understood that external sound turer's plant either, so there is a question of how much noise
control usually can reduce the noise to acceptable levels with­ the machine really produces - under any condition.
out disturbing the internal design, and in most cases can pro­
vide more noise reduction than is attainable by internal design Octave band sound pressure levels "under free field con­
changes alone. ditions" can be estimated with reasonable accuracy by applying
certain correction factors to the measured levels, as shown
External noise control techniques, such as acoustic enclo­ below:
sures, lagging, muffiers, vibration isolation, and damping,
must not be considered as "second class" noise reduction, but l. Measure octave band sound pressure levels according
as a perfectly acceptable approach. to ANSI SS.l-1971 (2), Test Code for the Measurement of
Sound from Pneumatic Equipment. This Code states that, in
the case of stationary equipment, measurements shall be made
Noise emission ratings are necessary
at each end of the equipment, and at the centers of the sides of
The purpose of sound specifications and sound guarantees each casing.
is to be sure you will not be in violation of some noise emission
2. At each microphone location, move the microphone
standard, or some noise immission regulation. Noise emission
away from the machine, in a direction perpendicular to the axis
standards set maximum sound pressure levels or sound power
of the machine, and note the maximum drop-off in sound pres­
levels with respect to the machine. Noise immission regu­
sure level that can be obtained, in each octave band of interest.
lations establish permissible sound levels in work locations, on
construction sites, or crossing plant boundaries or residential 3. At each microphone location note the distance from the
property lines. The OSHA Occupational Noise Exposure Reg­ machine where the maximum drop-off occurred, in each octave
ulation is an example of a noise immission regulation. band of interest. (That is, the distance beyond which no further
decrease in sound pressure level occurred.)
If a purchaser is interested only in complying with the
OSHA regulation, a statement of overall, A-Weighted sound 4. From Figure 2, determine the Correction Factor (1)
level in dBA is all that is needed. This is sufficient also if the to be subtracted from the measured octave band sound pres­
noise crossing the plant boundary is to be calculated and only sure level to obtain the approximate free-field level.
one machine is in operation.
5. Calculate the overall A-Weighted sound level (dBA) by
If there are a number of machines of different types, and applying the A-Network Corrections to each of the corrected
boundary sound levels are to be predicted, octave band data free-field levels above, and then combining the octave band
should be available, even if the property line levels are stated components.
 TURBOMACHINERY NOISE RATING 65

OCTAVf PASS BANDS IN CYClES PER SECOND

TWO DIFFERENT SOUNDS WITH THE SAME dBA READING Z10 - uoo - 2800

OCTAVE PASS BANDS (Hz)

0::
<(
III
0 11 0
0::
u
:i b-··-: t-·-
- 1>-·- �
1:'. -
/; - • _..;: ----- Indoor -
(IJ
0
0
0
100
�7 .
90 �--��-1�--+---�--����----I--�f---�- f'\Y:--· -
��
d � :--. -
w
0:: 90 F re; Field : ::
III 80 1-----=F-----=t�- -+---+- Corrected ·
.
..._ :
...."
"0
Indoor
!'!;
...J BO
w
>
w Ill
...J '"0
0
z 70
<(
III
w
>
� 60
u
0 100 1000 10000
FREQUENCY IN CYCLES PER SECOND

Figure 3. Machine No. 1 -Sound Pressure Levels Corrected

100 IK IOK to Free-Field Conditions.
FREQUENCY (Hz)

Figure 1. Two Different Sounds With the Same dBA Reading. 4. Therefore the approximate free-field sound pressure
level in the 500 Hz octave is:
10
1..- 92 dB - 2.5 dB = 89.5 dB

8 f---·
/ e�1-- 5. The equivalent overall dBA sound level may be calcu­
<If/ /" lated by subtracting the appropriate A-Weighting correction in

6
V/ jl1 1--t- each corrected octave band, and then combining the corrected
octave band levels.
// v--- 1 !'\ This procedure is not intended to give laboratory accu­

k:'-/�
�
racy, of course. It takes room constant into consideration, and
'2..0 makes a correction for it. It is certainly better than no correc­
1--- ----· -

v-
�
_; tion at all, and the method has been tested on small, medium

-- �
a.
(/) size, and large machines. Figures 3, 4 and 5, show typical test
I
..:r 3 ;j,
results.
8; v/ � r-- "'
Sometimes sound power level is best
0
2.5

�
-·

4.0 A-Weighted sound levels are all that are needed to deter­
...... mine compliance with the OSHA occupational noise exposure

��
4.5
-
regulation, and they can be measured quite simply at a
5 .0 machine operator's location. Not quite so simple, however, is
1.5
......

%
5.5 the prediction of noise contours within a plant or refinery area,
� e.o or the estimation of industrial plant or refinery boundary sound

� �
-
levels, when many different machines are in operation, at vari­
1 ous locations within the boundaries. Sound power levels are
1 2 3 4 5 6 8 9 10
preferred for this. The problem is determining the sound
DISTANCE r, METERS power level in the first place.
Figure 2. Correction Factors for Estimating Free-Field Sound
More care is required to determine sound power levels
Pressure Levels.
than to measure sound pressure levels. Sound power levels
must be calculated from measured sound pressure levels at a
EXAMPLE: number of specific locations around the machine, and the cal­
l. The sound pressure level at location 1, in the 500 Hz culated values can not be more accurate than the measured
octave band, is 92 dB, at a distance ofl meter from the machine. values. In fact, since additional steps are required in the calcu­
machine. lation process, there are more chances for error. In spite of this
sound power is a very useful quantity to have when making
2. As the microphone is moved away from the machine, acoustical calculations. It is practically independent of room
the sound pressure level in the 500 Hz band decreases to environment, and independent of distance from the machine.
89dB, at a distance of 3 meters from the surface of the machine. The procedure for determining it, though, is not independent
3. From Figure 2, the correction factor is found to be 2.5 of room constant, and that is where a major problem exists in
dB. practice. Because of its growing importance in noise control
66 PROCEEDINGS OF THE FIFTH TURBOMACHINERY SYMPOSIUM

OCTAVE PASS BANDS IN CYCLES PER SECOND OCTAVE PASS BANDS IN CYCLES PER SECOND

100

90 90

".E
,1! 80
�
� 80
'o 0
x x
q q
N N
� 70 � 70
ID ID
"0 "0

60

100 1000 10000 100 ,000 10000

FREQUENCY IN CYCLES PER SECOND FREQUENCY IN CYCLES PER SECOND
Figure 4. Machine No. 2- Sound Pressure Levels Corrected Figure 5. Machine No. 3 - Sound Pressure Levels Corrected
to Free-Field Conditions. to Free-Field Conditions.

regulations, both in the United States and Europe, considera­ waves from different parts of the machine come from nearly the
tion should be given to some of the factors involved in deter­ same direction, and the sound source can be treated as a "point
mining sound power in industrial environments, and the accu­ source."
racy that can be expected from the various procedures for ob­
4. If the measuring surface is a hemisphere over the floor,
taining it.
or reflecting surface, and if the hemisphere has a large radius,
Sound power determination the angle between the direction of sound wave propagation and
the surface vector is small, and the geometric error is negligi­
1. For accurate results, sound power level should be de­
ble.
termined in either a free field, an anechoic room, or a rever­
beration room. Large stationary machinery cannot be tested 5. In general, accuracy increases with the number of
in any of these three environments, and must be tested either measurement locations. For a true non-directional source, one
in the manufacturer's plant or on the purchaser's property, measurement point is sufficient.
under conditions which are far from ideal. In fact some ma­
6. If the microphone is placed closer to the machine, the
chines are installed in locations where it is simply impossible to
number of measurement locations should be increased so that
conduct a sound test that has any meaning. Some machines
the distance between adjacent positions is not larger than the
are nearly as large as the room in which they are located. In
measurement distance from the machine.
other cases, other nearby equipment, which cannot be shut
down, makes more noise than the machine being tested. 7. In most industrial locations it is impossible to make
sound measurements on the surface of a hemisphere whose
In most cases the problem is not as difficult as it might
radius is two or three times the largest dimension of the
seem at first, and both octave band sound pressure levels and
machine. In fact, in indoor locations, it is usually not possible
octave band sound power levels can be obtained with reason­
to have the measurement distance even one times the machine
able accuracy.
dimension, and it is customary to make sound pressure level
2. The sound power level of a machine, or sound source, measurements at a distance of one meter from the machine.
must be calculated from sound pressure levels measured on an This means that microphones would have to be located not
imaginary closed surface around the source. This is shown by more than one meter apart, and this would result in a large
Equation 1: number of measurements.
s
Lw = Lp + 10 log -- (1) Several recent investigations have shown that reasonably
Sa accurate sound power level determinations can be made when
where Lw = Sound power level in decibels re 10-12 watt. using microphone distances of 0.3 to 1.0 meter from the
Lp Sound pressure level, in decibels re 20 micro­ machine (3,4,5,6).
pascals.
8. A number of Draft International Standards (7,8)
s = Area of the measurement surface in square recently developed by the International Organization for Stan­
meters. dardization, ISO, recommended that the measurement sur­
S0 = 1 square meter. face be in the form of a rectangular parallelepiped, at a dis­
tance of one meter from a reference surface, which is defined
3. When the measurement distance is large in compari­ as the smallest possible imaginary rectangular parallelepiped
son with the dimensions of a noise-radiating machine, all sound that will just enclose the machine under test.
 TURBOMACHINERY NOISE RUING 67

9. In equation (1) above, S0 is 1 m2, simply to keep the .- --- -------

d - --- - --�
ratio of S to S0 dimensionless. The sound power level, L, is
equal to the average sound pressure level on the measure­ I 1--------.!..------- -1 I
ment surface plus 10 log S. I 1 I

ild H lr!
For most work, the exact area of S is not too important.
An error of 25 per cent results in an error of only 1dB.

10. Although there is very little error in using the "box"

measurement surface in testing large machinery, there can be
some problems in the case of small machines, where the error ,---------�--------,
d

in measurement area can be much greater than 25 per cent. , - - - -, -

I
- --------- - -

In fact the United States recently voted disapproval of the I

Draft International Standard for pneumatic tools and rock I I I
I
.al•l
drills (7) because of uncertainty concerning the proper mea­
surement area. This could vary as much as 318 per cent, and
produce an error of more than 5dB.
l I
, ,
I I
11. Even in the case of large machines, a rectangular
I I
_________________

parallelepiped has certain deficiencies as a measurement sur­ �-- - -- --- ----- -----�
face. Not all points are equidistant from the reference surface; Figure 6. Two-Surface Method Measurement Swfaces.
the corners of the box are farther than 1 meter away. Instead of
a sharp cornered box, one with rounded corners should be
14. In the "'two-surface method" octave band sound pres­
used, so that it conforms to a Huygens' surface. Huygens' prin­
sure levels are measured on the surfaces of two hypothetical
ciple states that all points on a vibrating surface may be consid­
rectangular parallelepipeds over the machine being tested.
ered to be the center of a spherical wave whose intensity is
proportional to the excitation at that point. The second "box" is farther from the machine than the
first box, and therefore its area S2 is greater than the area of
The justification for the sharp cornered box is that it is
the first, S1, as shown on Figure 6.
very nearly correct, it is much easier to calculate, and further­
more, sound pressure levels measured at the corners should be The average sound pressure levels on area S1 will be
lower than at other points, because they are farther away, and greater than the average of the levels on S2, because S1 is
thereby partially compensate for the increased area. closer to the machine than area s2.

Unfortunately some test codes do not require sound pres­ 15. The equations relating sound power level, sound
sure measurements at the corners. Recent ISO standards do pressure level, and room constant, for the two parallelepipeds

(l; + �
require this, however. are:

12. A calculation, or an estimate, of the room constant is Lw = L p1 10 log (2)

(t + �)
necessary in order to calculate sound power level from mea­
sured sound pressure levels. This can be done in several ways: Lw = Lp2- 10log
(a) It can be calculated by estimating the absorption coef­
ficients of the floor, ceiling, side walls, and other items in the where Lw is the sound power level in dB re 10"12 watt.
room, and estimating the individual areas of each. It is obvious
that this technique is not very accurate for industrial plant
Lp1 is the average sound pressure level on surface S1,
re 20 micro-pascals.
locations.
Lp2 is the average sound pressure level on surface S2
(b) It can be determined experimentally by measuring the re 20 micro-pascals.
reverberation time with a microphone, sound level meter, and R is the room constant in square meters.
high speed graphic level recorder. A wide band noise source is
stopped suddenly, and the time is measured, in each octave R can be eliminated from these two equations and sound
band, or third octave band, for the level to decrease 60 dec­ power level can be shown to equal
ibels. Lw = Lp + 10 log St - C (3)
(c) A comparison test can be made with another sound
source. A laboratory calibrated reference source with known
octave band, or third octave band sound power levels can be
where C = 10log�- (1- 81
K- 1 \ s2
)
used in an absolute comparison test, or an auxiliary sound
source can be used in a relative comparison test. In both of 10(Lp1 LP2)/10
-

and K =

these procedures the machine under investigation should be

moved out while measurements are being made on the refer­ 16. The environmental correction C, may be calculated
ence source. This is obviously impossible, in the case of large mathematically; or it may be found more conveniently from
machinery, and the technique becomes questionable when it is a set of curves, Figure 7 relating the correction factor to
considered that many different answers are obtained when the Lp1 - Lp2 and the area ratio S1 /S2.
reference source is placed at various locations with respect to
Verification of two-suiface method
the machine under test.
The Diesel Engine Manufacturers Association, D EMA,
13. One of the best techniques is to use the machine recently conducted a series of sound tests to compare sound
under test as its own "reference source" to evaluate the acous­ power levels determined by the two-surface method with those
tic environment. This is the basis of the "two-surface method." obtained using reverberation time procedures.
68 PROCEEDINGS OF THE FIFTH TURBOMACHINERY SYMPOSIUM

10
9
Accuracy of results
8

7
c� ...... A guide to the accuracy that can be expected in determin­
� ......
I.Q ...... ing sound power levels is given below. This should be consid­
5
Ill
......___ ered whenever it becomes necessary to verifY sound estimates
..,
r-.1-5 ...... '·""
or guarantees.
� ...... � �
-

li
I
3
-
a?- ...... ....... ,'\1\
......
ISO Draft International Standard 3989, covering the mea­
� - ...... '0 \: �
5
surement of airborne noise emitted by compressor-prime
I� 2
mover units, states that in a free-field over a reflecting plane
-%
r--ii- -
:::: ...... � � �� outdoors, where there are no large interfering surfaces; or in
:::: .. �
� ��
5

� ::::: -....
s.'s"
...... �
indoor locations where the effect of the environment can be
1
.015 .02 .025 .03 .04 .05 .oe .aT .o6.01UO .15 2 .2 5 3 4 .5 6 7 .8 9 tO
established by either a relative comparison test or a reverbera­
.01 0.1 1.0
tion test, the uncertainty in determining sound power levels is
SdSo AREA RATIO
as follows:
Figure 7. Two-Surface Method Correction Factors. Octave-Band Standard
Center Frequency Deviation
INGERSOLL- RAND DIEHL- TWO SURFACE METHOD Hz dB
ENGINE Q REVERBERATION TIM£ METHOD

63 5.0 (Approx.)
11 WHISTLE DISTURBANCE
125 3. 0
250 2.0
WHISTLE DISTURBANCE
K
500 2.0
lK 1.5
2K 1.5
4K 1.5
8K 2.5

These are the standard deviations to be expected under

31.5 03 us 250 soo 1,000 apoo 4.ooo a.ooo te.ooo near ideal conditions, and do not include variations in the
sound power level of the machine under test.
OCTAVE lAND CENTER FR!GUENCY

Figure 8. Sound Power Level Comparison - Two-Surface

ISO/DIS 3989 points out that in practical industrial envi­
Method vs Reverberation Time Method.
ronments "the accuracy stated in the body of this International
Standard can not be assured."
COOPER•BE88EIIER ENGINE
0 DIEHL-TWO SURFM:E IIETMOO
Q REVER8ERATION TillE METHOD What this means is that machinery manufacturers, pur­
chasers, and those who write sound specifications for machin­
ery must take a realistic approach in defining requirements of
sound guarantees and in making measurements to verifY them.

Figures 8 and 9 show that, at least in these tests, the

two-surface method offers a practical procedure for sound
power determination, and provides reasonable accuracy- suf­
ficient even for regulatory purposes, provided the regulations
also take a realistic approach.

Conclusions and recommendations

1. Sound specifications should be written \\ith a definite
31.5 13 125 250 SOO 1.000 2.000 4.000 a.ooo 1e,aoo objective in mind. This should not be simply to purchase qui­
OCTAVE llANO CI!NTIR FREQUENCY
eter machinery.
Figure 9. Sound Power Level Comparison - Two-Surface For example, the purpose of the specification may be to
Method vs Reverberation Time Method. assure compliance with OSHA regulations at worker locations;
or it may be to meet an EPA product noise emission standard
Two different diesel engines were tested in this investiga­ so that the machines can be released for distribution into com­
tion, one manufactured by Ingersoll-Rand Company and one merce; or it may be to comply with a State or local noise control
by Cooper-Bessemer Company. The results of the tests will ordinance stating maximum boundary sound levels.
probably be published later in a technical paper by DEMA. 2. It is a costly mistake to demand more noise control
Figures 8 and 9, taken from the data show that the two than is necessary.
methods agree remarkably well. 3. It takes more time, and therefore more money, to get
Sound pressure levels must be measured in any sound octave band data than it does to get overall dBA data but in
power determination - no matter which technique is used. many cases octave band levels are necessary.
The two-surface method requires a minimum of sound test 4. Sound guarantees in terms of octave band levels are
instrumentation, and can be performed without removing the much more stringent than those for overall levels only, even
machine from its installation, and even without shutting it though the overall levels are identical in both cases. The
down. It appears to be the best of the practical methods for guarantee must be met in eight, or nine octaves, instead of at
industrial environments. one point.
 TURBOMACHINERY NOISE R\TING 69

5. In general, sound power data is more valuable to have Sound power level - Sound power level, in decibels, is 10 times
than sound pressure information. It is also more costly to ob­ the logarithm, to the base 10, of the ratio of a given power to a
tain. It requires more microphone locations, more data to pro­ reference power (usually one pico-watt).
cess, and more calculations than sound pressure level data.
Sound pressure level - Sound pressure level, in decibels, of a
6. The two surface method is one of the best techniques sound is 20 times the logarithm, to the base 10, of the ratio of
to use for sound power determinations in industrial locations. the pressure of the sound to a reference pressure (usually 20
micro-pascals).
7. The true accuracy of any sound power determination
made on large machinery in industrial locations is almost im­ Third octave band- A third octave is the interval between two
possible to obtain, because there is no way to determine the sounds having a basic frequency ratio equal to the cube root
true accuracy for comparison purposes. In general, accuracy of2.
should be within 3 to 5 decibels of the true value, if sound tests
Two-surface method - A method for determining sound power
are conducted properly.
level from sound pressure levels measured on the surfaces of
8. Machinery manufacturers, purchasers, and legislators two hypothetical parallelepipeds enclosing the machine under
must take a realistic approach in establishing maximum per­ test.
missible sound levels.

APPENDIX A REFERENCES
Definitions 1. George M. Diehl, "Machinery Acoustics," published by
Wiley-lnterscience, New York, N.Y.
Absorption coefficient - The sound absorption coefficient of a
surface is the fraction of incident sound energy absorbed or 2. ANSI S5.1-1971 "Test Code for the Measurement of
otherwise not reflected by the surface. Sound from Pneumatic Equipment," American National
Anechoic room - An anechoic room is one whose boundaries Standards Institute, New York, N.Y.
absorb effectively all the sound incident thereon, thereby af­
3. Holmer, C.l., "Procedures for estimating sound power
fording essentially free-field conditions.
from measurements of sound pressure: An experimental
A-Weighting - Sound level measurements made with the sound investigation with application to noise from portable air
level meter set on the A-scale are said to have A-Weighting. compressors." National Bureau of Standards, July 1975.
This electrical network, which discriminates against low fre­ Available from National Technical Information Service,
quencies, has a frequency response similar to that of the Springfield, Va. 22151 - Accession No. COM-75-11399.
human ear.
4. 0. L. Angevine, "Nearfield Measurement and Sound
Decibel - A decibel is the unit of level when the base of the Power." Unpublished paper presented at 83rd meeting of
logarithm is the tenth root of 10, and the quantities concerned ASA, April 1972.
are proportional to power.
5. 0. L. Angevine and R. F. O'Donnell, "A Method for De­
Free field A free sound field is a field in a homogeneous,
-
termining Sound Power Levels of Large Machinery." Un­
isotropic medium free from boundaries. In practice it is a field published paper presented at the 87th meeting of ASA,
in which the effects of the boundaries are negligible over the April 1974.
region of interest.
6. Hubner, G., "Analysis of errors in measuring machine
Octave band - An octave is the interval between two sounds
noise under free-field conditions." Journal A coustical
having a basic frequency ratio of two.
Society of America, 54-1973.
Reverberation room - A reverberation room is a room having a
long reverbation time, especially designed to make the sound 7. ISO/DIS 3481 Acoustics - Measurement of airborne
field therein as diffuse as possible. noise emitted by pneumatic tools and machines - En­
gineering method for determination of sound power
Room constant - Room constant is equal to the product of the levels.
average absorption coefficient of the room and the total inter­
nal area of the room divided by the quantity one minus the 8. ISO/DIS 3989 Acoustics - Measurement of airborne
average absorption coefficient. noise emitted by compressor units including prime
movers - Engineering method for determination of
If there is a large amount of absorption present, the room
sound power levels.
is said to have a large room constant, and it behaves in a
manner similar to a free field. If there is only a small amount of 9. George M. Diehl, "Sound Power Measurements on Large
absorption, the room has a small room constant, and its charac­ Machin ery Installed Indoors" - Compressed Air,
teristics are nearer to those of a reverberant Toom. January, 1974.
70 PROCEEDINGS OF THE FIFTH TURBOMACHINERY SYMPOSIUM