You are on page 1of 15

Evaluated Uncertainties for Measurements

of Airborne Sound Attenuation between


Rooms in Buildings
by

Hoda S. Seddeq

Reprinted from

JOURNAL OF
BUILDING ACOUSTICS
Volume 17 · Number 1 · 2010

MULTI-SCIENCE PUBLISHING CO. LTD.


5 Wates Way, Brentwood, Essex CM15 9TB, United Kingdom
BUILDING ACOUSTICS · Volume 17 · Number 1 · 2010 Pages 69–82 69

Evaluated Uncertainties for Measurements


of Airborne Sound Attenuation between
Rooms in Buildings
Hoda S. Seddeq*
Department of Building Physics & Environment
Housing & Building National Research Centre,
Cairo,-Egypt

(Received 1 September 2009 and accepted 17 February 2010)

ABSTRACT
The standard ISO/IEC 17025:2005 “General requirements for the competence of testing and
calibration laboratories” introduces the necessity of evaluation of measurement uncertainties for
making them reliable and comparable. In order to fulfill acoustic comfort conditions indoors,
building acoustic have to establishes a set of criteria for the acoustic performance of the
building. These criteria may be expressed in terms of permissible minimum values for
normalized noise reduction NNR, noise reduction NR or apparent transmission loss ATL. This
paper discusses the uncertainty sources and presents a methodology for the calculation of
uncertainty of the airborne sound insulation between rooms carried out according to the of
standards ASTME 336 and examples of uncertainty budget, evaluated for the measurement
results of the sound insulation. The range of values obtained for the measurement results
uncertainty is 0.68 to 1.77.

Keywords: uncertainty budgets, combined standard uncertainty, sensitivity coefficients,


expanded uncertainty, normalized noise reduction, apparent transmission loss, noise reduction

1. INTRODUCTION
For noise control in buildings a set of criteria for the acoustic performance of building
has to be established in order to fulfill acoustic comfort conditions indoors. In the case
of a field evaluation of airborne sound insulation of a separating element between two
rooms, the uncertainty must be evaluated and added to the final result to be compared
to the prescribed limit. The housing and building national research centre in Egypt are
encouraged that tests in the field of building physics should have accreditation
according to the standard ISO/IEC 17025 [1]. Therefore the building acoustics
department in the housing and building national research centre should perform a
complete evaluation of uncertainty of in-situ sound insulation measurements.
70 Evaluated Uncertainties for Measurements of Airborne Sound Attenuation
between Rooms in Buildings

The standard ASTM E 336 uses a procedure originally developed for laboratory
measurements of the transmission loss of partitions. These procedures assume that the
rooms in which the measurement is made have a sound field that reasonably
approximates a diffuse field. Sound pressure levels in such rooms are reasonably
uniform throughout the room and average levels vary inversely with the logarithm of
the room sound absorption. Unfortunately not all rooms not specifically designed as
laboratory testing rooms will satisfy these conditions and conditions in a field test
should not be modified by any temporary means to improve performance of testing.
Practical experience and studies have shown that the test method is applicable to
smaller spaces with volume less than 150 cubic meters. The repeatability for this test
method may be due to variation in microphone and loudspeaker positions [2].
Higginson, R. F. estimated the repeatability standard deviations of the same
measurement team between the same two rooms using the same equipments as
follows: 2 dB for the frequency bands 100 to 200 Hz and 1 dB for the bands above
that. The corresponding 95% repeatability limits are 5.6 dB and 2.8 dB [3]. ASTM E
90 prescribes procedure for determining where the sound fields in the rooms and
microphone systems used to sample them give measurements uncertainties that are
low enough for the purpose of this test method. This is decided by calculating
confidence intervals from measurements of sound pressure level and reverberation
time [4].
The strict mathematical way for estimating uncertainties is described extensively in
the GUM [5], The uncertainty of measurement associated with the input estimates is
evaluated according to either a ‘Type A’ or a ‘Type B’ method of evaluation. The Type
A evaluation of standard uncertainty is the method of evaluating the uncertainty by
the statistical analysis of a series of observations. In this case the standard uncertainty
is the experimental standard deviation of the mean that follows from an averaging
procedure or an appropriate regression analysis. The Type B evaluation of standard
uncertainty is the method of evaluating the uncertainty by means other than the
statistical analysis of a series of observations. In this case the evaluation of the standard
uncertainty is based on some other scientific knowledge [5].

2. CALCULATION OF THE STANDARD UNCERTAINTY


The standard uncertainty, u(xi ) can be obtained when several independent observations
have been made for one of the input quantities under the same conditions of
measurement. If there is sufficient resolution in the measurement process there will be
an observable scatter or spread in the values obtained in such case the standard
deviation will be used directly as standard uncertainty as:

u( x i ) = s ( x i ) (1)

where s( x ): experimental standard deviation of the mean for n independent


measurements.
BUILDING ACOUSTICS · Volume 17 · Number 1 · 2010 71

The standard deviation of the estimated measurement result y (called combined


standard uncertainty uc(y) is obtained by combining the individual standard
uncertainties u(xi) using the usual method of combining standard deviations
2
n
 ∂f 
uc 2 ( y) = ∑   u 2 ( xi ) (2)
i =1  ∂x i 

 ∂f 
where the partial derivative,  is termed a sensitivity coefficient.
 ∂x i 

The interval in which the value of the quantity subject to measurement can be asserted
to lie with a specified level of confidence is referred to as the, expanded uncertainty U
(it means that one can with the relevant level of confidence believe that the value of Y,
estimated by y, lies within the limits y – U ≤ Y ≤ y + U ). The expanded uncertainty is
obtained by multiplying a combined standard uncertainty by a coverage factor k
depending on the desired level of confidence and the type of statistical distribution.

U(Y) = k uc (y) (3)

In cases where a normal (Gaussian) distribution can be attributed to the measurand


and the standard uncertainty associated with the output estimate has sufficient
reliability, the standard coverage factor is k = 2 and the assigned expanded uncertainty
corresponds to a coverage probability of approximately 95% [6], [7].
For Rectangular distributios upper and lower limits (a- and a+) (example: last digit
of a digital display), are used to determine an uncertainty, u(xi), for that component [6]:

(a− − a+ ) / 2
u( x i ) = (4)
3

2.1. Uncertainty components for normalized noise reduction measurements


Where both rooms are less than 150 cubic meters, the normalized noise reduction, NNR
in certain frequency band shall be calculated as the difference between the average
sound pressure levels obtained in the source and receiving rooms using [2]:

T
NNR = LS − LR + 10 log (5)
0.5
where
LS = the average sound pressure level in the source room, dB, and
LR = the average sound pressure level in the receiving room, dB,
T = the average reverberation time in the receiving room,
72 Evaluated Uncertainties for Measurements of Airborne Sound Attenuation
between Rooms in Buildings

For each frequency band equation (5) can be simplified to:

NNR = L S − LR + 10 log T + 3 (6)

For each frequency band the averaged sound pressure L in each room: is calculated
according to:

1 n Li

L = 10 log  ∑ 10 10  (7)
 n i =1 

Each individual sound pressure level measurement, Li, will contribute to the type A
uncertainty. The standard uncertainty is calculated from the standard deviation, s(Li) for
N different measurements as follows:

u( L ) = s( Li ) / N (8)

The contributions to the total uncertainty are summarized in table 1


where:
• u(LS): the uncertainty due to the sound pressure level measurement in the source
room
• u(LR): the uncertainty due to the sound pressure level measurement in the
receiving room
• u(T): the uncertainty due to the reverberation time measurement
• δLana: the correction associated with the calibration of the building acoustic
analyzer
• u(δLana): the uncertainty due to building acoustic analyzer calibration
• U(δLana): the expanded uncertainty for building acoustic analyzer calibration
• δLmic: the correction associated with the calibration of microphone
• u(δLmic): the uncertainty due to microphone calibration
• U(δLmic): the expanded uncertainty for microphone calibration
• δLcal: the correction associated with the calibration of the calibration checker
• u(δLcal): uncertainty due to the calibrator calibration
• U(δLcal): the expanded uncertainty for the used calibrator calibration
• δ Lpramp is the correction associated with the calibration of the preamplifier
• u(δLpramp): uncertainty due to the preamplifier calibration
• U(δLpramp): the expanded uncertainty for the preamplifier calibration
• δ Tres: is the correction associated with the resolution of the building acoustic
analyzer display
• u(δ Tres): the uncertainty due to the resolution of the building acoustic analyzer
display
BUILDING ACOUSTICS · Volume 17 · Number 1 · 2010 73

Table 1: Uncertainty budget for the normalized noise reduction measurements

Standard Uncertainty
Quantity Uncertainty Source of Probability uncertainty contribution
Xi symbol uncertainty distribution u(xi) ui(y) = ciu(xi)

LS u(Ls) Repeatability of Normal s( LSi ) s( LSi )


LS for N n n
measurements
LR u(LR ) Repeatability of Normal s( LRi ) s( LRi )

LR for N n n
measurements
T u(T ) Repeatability of Normal s(Ti )  4.34   s(Ti ) 
T for N n  T   n 
measurements
δLana u(δLana) Building Normal U (δ Lana ) U (δ Lana )
acoustic analyzer kN ,95% kN ,95%
calibration
δLmic u(δLmic) Microphone Normal U (δ Lmic ) U (δ Lmic )
calibration kN ,95% kN ,95%
δLcal u(δLcal) The correction Normal U ( Lcal . ) U ( Lcal . )
for the calibrator kN ,95% kN ,95%
δLpramp u(δLpramp.) preamplifier Normal U (δ L pramp. ) U (δ L preamp. )
calibration kN ,95% kN ,95%

δTres u(δTres) Resolution of Rectangular Tres  4.34   Tres 


reverberation 2 3  T   2 3 
time

• kN, 95%: the coverage factor for confidence level 95%


• The sensitivity coefficients, ci are:

∂NNR ∂NNR ∂NNR 4.343


= 1, = −1, =
∂LS ∂LR ∂T T

The analysis of the uncertainty budget is based on the following assumptions:


• the distributions are normal and the confidence level is equal to 95% (coverage
factor equal 2) for standard uncertainties of LS, LR, T
• uncertainties due to δLana, δLmic, δLcal, δLpramp assume calibration certificates
with normal distribution and confidence level of 95%-
74 Evaluated Uncertainties for Measurements of Airborne Sound Attenuation
between Rooms in Buildings

2.2. Uncertainty components for apparent transmission loss measurement


The apparent transmission loss, ATL is calculated from the difference between the
average sound pressure levels obtained in the source and receiving rooms using [2]:
 S
ATL = LS − LR + 10 log   (9)
 A2 

where
LS = the average sound pressure level in the source room, dB, and
LR = the average sound pressure level in the receiving room, dB, and
S = the area of the tested sample, m2, and
A2 = the sound absorption in the receiving room m2
The sound absorption of the receiving room, A2 is determined, by [8]:

A2 = 0.921 Vd/c (10)

where:
c = speed of sound in air, m/s,
V = volume of room, m3, and
d = rate of decay of sound pressure level in the room, dB/s.
where:
d = 60/T,
T = average reverberation time in the receiving room, s.
The speed of sound changes with temperature, and is calculated for the conditions
existing at the time of test from the equation [8]:
1
c = 20.047 ( 273.15 + t ) 2 m/s (11)

where:
t = receiving room temperature, °C.
For very large halls with highly reflecting surfaces, air absorption at high frequencies
can be the dominant phenomenon which is strongly influenced by the relative humidity.
For frequencies below 2 kHz, absorption due to the air is not significant and therefore
ignored [9].
For each frequency band equation (9) can be expanded as :

ATL = LS − LR + 10 log s + 10 log T − 10 log V + 5 log(2273.15 + t ) − 4.4 (12)

The contributions to the uncertainty are summarized in table 2.


where
• u(t): the uncertainty due to the temperature measurement
• δtcal: the correction associated with the calibration of the thermometer
BUILDING ACOUSTICS · Volume 17 · Number 1 · 2010 75

Table 2: Uncertainty budget for the apparent transmission loss measurements


for each

Standard Uncertainty
Quantity Uncertainty Source of Probability Uncertainty contribution
Xi symbol uncertainty Distribution u(xi) ui(y) = ci u(xi)

LS u(Ls) Repeatability of Normal s( LSi ) s( LSi )


level, LS for N n n
measurements
LR u(LR) Repeatability of Normal s( LRi ) s( LRi )

level, LR for N n n
measurements
T u(T) Repeatability of Normal s(Ti )  4.34   s(Ti ) 
reverberation, T for n  T   n 
N measurements
S u(S) Repeatability of Normal s ( Si )  4.34   s(Si ) 
area, S for N n  S   n 
measurements
4.34   s(Vi ) 
− 
V u(V) Repeatability of Normal s(Vi )
volume, V for N n  V   n 
measurements
t u(t) Repeatability of Normal s ( ti )  2.17  s(ti )
 
temperature, t for N n 273.15 + t  n
measurements
δLana u(δLana) Building acoustic Normal U (δ Lana ) U (δ Lana )
analyzer kN ,95% kN ,95%
calibration
δLmic u(δLmic) Microphone Normal U (δ Lmic ) U (δ Lmic )
calibration kN ,95% kN ,95%
δLpramp u(δLpramp.) Preamplifier Normal U (δ L pramp. ) U (δ L preamp.. )
calibration kN ,95% kN ,95%

δLcal u(δLcal.) Correction for Normal U (δ Lcal . ) U (δ Lcal . )


the calibrator kN ,95% kN ,95%

δScal u(δScal) Calibration Normal U (δ Scal . ) 4.343 U (δ Scal . )


of area kN ,95% S kN ,95%

δVcal u(δVcal) Calibration of Normal U (δ Vcal . ) 4.343 U (Vcal . )



volume kN ,95% V kN ,95%

δtcal u(δ tca.) Temperature Normal U (tCal . ) 2.17 U (t cal . )


calibration kN ,95% 273.15 + t kN ,95%

δTres u(δ Tres.) Resolution of Rectangular Tres  4.34   Tres 


reverberation 2 3  T   2 3 
time reading
76 Evaluated Uncertainties for Measurements of Airborne Sound Attenuation
between Rooms in Buildings

• u(δtcal): uncertainty due to thermometer calibration


• U(δtcal): the expanded uncertainty for temperature calibration
• u(V): the uncertainty due to the volume measurements
• δVcal : the correction associated with calibration for volume measurements
• u(δVcal): uncertainty due to volume calibration
• U(δVcal): the expanded uncertainty for volume calibration
• u(S): the uncertainty due to the area measurements
• δScal: the correction associated with the calibration for area measurements
• u(δScal): uncertainty due to calibration for area measurements
• U(δScal): the expanded uncertainty for area calibration
• The sensitivity coefficients are equal to:

∂ATL ∂ATL ∂ATL 4.343 ∂ATL 4.343 ∂ATL 4.343


= 1, = −1, =− , = , =
∂LS ∂LR ∂V V ∂S S ∂T T
∂ATL 2.17
=
∂t 273.15 + t

The analysis of the uncertainty budget assumes the following :


• uncertainties of the volume, area and temperature calibration correction δVcal,
δScal, δtcal, are evaluated on the basis of calibration certificates having a normal
distribution and confidence level 95%-.
• the distribution is normal and the confidence level is equal to 95% for standard
uncertainties of the estimated quantities V and S.

2.3. Uncertainty components for noise reduction measurements


The noise reduction in certain frequency band shall be calculated as the difference
between the average sound pressure levels obtained in the source and receiving rooms
using [2]:

NR = LS − LR (13)

where
LS = the average sound pressure level in the source room, dB, and
LR = the average sound pressure level in the receiving room, dB.
where the sensitivity coefficients, ci are equal to:

∂NR ∂NR
=1, = −1
∂LS ∂LR

Table 3 summarized the uncertainty budget for noise reduction measurements.


BUILDING ACOUSTICS · Volume 17 · Number 1 · 2010 77

Table 3: Uncertainty budget for the noise reduction measurements

Standard Uncertainty
Quantity Uncertainty Source of Probability Uncertainty contribution
Xi symbol uncertainty Distribution u(xi) ui(y) = ci u(xi)

LS u(LS) Repeatability Normal s( LSi ) s( LSi )


of LS for N n n
measurements
LR u(LR) Repeatability Normal s( LRi ) s( LRi )

of LR for N n n
measurements
δLana u(δLana) Building Normal U (δ Lana ) U (δ Lana )
acoustic kN ,95% kN ,95%
analyzer
calibration
δLmic u(δLmic) Microphone Normal U (δ Lmic ) U (δ Lmic )
calibration kN ,95% kN ,95%

δLcal u(δLcal.) Correction Normal U (δ Lcal . ) U (δ Lcal . )


for the kN ,95% kN ,95%
calibrator
δLpramp u(δLpream.) Preamplifier Normal U (δ L preamp ) U (δ L preamp. )
calibration
kN ,95% kN ,95%

3. MEASUREMENT PROCEDURES AND EXPERIMENTAL RESULTS


The airborne sound insulation between two adjacent rooms was measured from 100 Hz
to 4000 Hz according to ASTM E336. The source room volume is 65.2 m3 and the
receiving room volume is 52 m3. Five microphone positions and two source positions
were used. The sound signals used for the tests are pink noise over third octave bands
with mid-band frequencies from 100 to 8000 Hz. The noise signal is radiated by the
omniloudspeaker type 4229 (B&K) connected with power amplifier type 2716 (B&K).
The omniloudspeaker was far enough away from the test partition and radiated levels
adequately above the background noise (more than 10 dB). A 1/2 inch fixed microphone
type 4189 (B&K) connected with the building acoustic analyzer type 4417 (B&K) was
used to measure average sound pressure levels in the rooms and sound decay rates in
the receiving room. In the measurement of reverberation time, 5 samples were taken for
each frequency band and the reverberation time was deduced from the average of the
decay curve of the samples according to ASTM E2235 [8]. The tests were repeated
three independent times, calibrating the measuring system before each measurement
78 Evaluated Uncertainties for Measurements of Airborne Sound Attenuation
between Rooms in Buildings

70
Transmission loss and level difference (dB)

60

50

40

30
ATL
NR
20 NNR

10
100 125 160 200 250 315 400 500 630 800 1000 1250 1600 2000 2500 3150 4000
Third octave frequencies Hz

Figure 1. Mean values of the apparent transmission loss ATL, normalized noise
reduction NNR and noise reduction NR.

with a 1000 Hz sound level calibrator type 4231 (B&K). The reverberation time of the
receiving room was measured with two source positions and five microphone positions,
two measurements each. The area of the separating wall was 13 m2.
Figure 1 presents the mean values of the apparent transmission loss ATL, normalized
noise reduction, NNR and noise reduction, NR.

3.1. Results
Figures (2), (3), (4) show the contribution for sound insulation measurement results at
1000 Hz. These figures show that the sound pressure level measurements make the
most contribution to the total uncertainty. The measurement of reverberation time also
has a significant effect on the total uncertainty. Calibration errors – as may be expected
– made less contribution to the total uncertainty. Figure (4) shows the contribution of
measuring area and volume is low. The contribution of temperature measurement is
very low and may be neglected.
Table (4) gives the values of the expanded uncertainty, with 95% confidence
interval (k = 2) for the measured ATL, NNR and NR at third octave frequencies from
100 to 4000 Hz. These values obtained for the expanded uncertainty range between
BUILDING ACOUSTICS · Volume 17 · Number 1 · 2010 79

1
0.9

0.8
Uncertainty contribution value
0.7

0.6

0.5

0.4

0.3
0.2

0.1

0
)

T)

p)

l)

uc
LS

LR

ic

na

es
ca
m
u(

Tr
a
u(

u(


ra

δL

δL

u(
p

u(
u(

u(
δL
u(

Figure 2. Uncertainty contributions in NNR results at 1000 Hz.

0.9

0.8
Uncertainty contribution value

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
u(LS) u(LR) u(δLpramp) u(δLmic) u(δLana) u(δLcal) uc

Figure 3. Uncertainty contributions in NR results at 1000 Hz.


80 Evaluated Uncertainties for Measurements of Airborne Sound Attenuation
between Rooms in Buildings

0.9

0.8
Uncertainty contribution value

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
)

v)

s)

t)

T)

p)

)
.)

u( l)

)
uc
LS

LR

al

ic

na

es

al
ca
u(

al
u(

u(

am
u(

m
c

Vc
Sc

Tr
a
u(

u(

δt


δL

δL
pr

u(
u(

u(

u(
u(

u(
δL
u(

Figure 4. Uncertainty contribution in ATL results at 1000 Hz.

Table 4: Expanded uncertainty with 95% confidence


interval (k = 2) for the measured ATL, NNR and NR at
third octave frequencies from 100 to 4000 Hz

Freq(Hz) U (ATL) U (NR) U (NNR)

100 1.482 1.478 1.482


125 1.335 1.321 1.335
160 1.497 1.489 1.497
200 1.637 1.636 1.637
250 1.750 1.745 1.750
315 1.002 1.000 1.002
400 1.581 1.143 1.580
500 1.302 1.234 1.302
630 1.637 1.401 1.636
800 1.068 1.035 1.068
1000 1.768 1.757 1.767
1250 1.250 1.249 1.250
1600 1.757 1.387 1.757
2000 1.321 1.318 1.321
2500 1.376 1.292 1.376
3150 0.984 0.978 0.984
4000 0.679 0.677 0.679
BUILDING ACOUSTICS · Volume 17 · Number 1 · 2010 81

0.68 and 1.77. These fall within the limits published in ASTM E336 and Portuguese
legislation [10].

4. CONCLUSION
This study presents a methodology for the standard uncertainty calculation of the field
airborne sound insulation measured according to ASTM E336. The analysis of
uncertainties was carried out using the same measurement team making measurements
between the same two rooms using the same equipment. The rooms were small spaces
with volumes less than 150 cubic meters. The final combined uncertainty for airborne
sound attenuation was found to depend mainly on the spread of values of sound level
in the source room, sound level in the receiving room and the reverberation time. To
reduce the magnitude of the combined total standard uncertainty effort should be
concentrated on reducing the variation of sound pressure level in rooms and variation
of reverberation time by selecting suitable positions of the sound source and
microphone. Whilst it is clear that the magnitude of the final combined total standard
uncertainty effort is affected by calibration errors of the instrumentation this was minor
in comparison.
The area and volume measurement made a quite low contribution to the total
uncertainty. The temperature uncertainty had a negligible effect on the total uncertainty
so it can be neglected.

REFERENCES
[1] ISO/IEC 17025, General requirements for the competence of testing and
calibration laboratories, Second edition, 2005-05-15
[2] ASTM E336, Standard test method for measurement of airborne sound
attenuation between rooms in buildings, 2005
[3] Higginson, R. F., A Study of Measuring Techniques for Airborne Sound Insulation
in Building” Journal of Sound and Vibration, Vol. 21, 1972 p. 405
[4] ASTM E90-04, Standard test method for laboratory measurement of airborne
sound transmission loss of building partition and elements
[5] ISO Guide 98, Guide to the Expression of Uncertainty in Measurement.
International Organization for Standardization, 1995
[6] EA-4/02, Expression of the Uncertainty of Measurement in Calibration, European
cooperation for Accreditation, December, 1999.
[7] EUROLAB Technical Report 1/2002 , Measurement Uncertainty in testing, 2002
[8] ASTM E2235, Test Method For The Measurement of Decay Rates for Use In
Sound Insulation Test Methods
[9] Dennis A. Bohn, Environmental Effects on the Speed of Sound, J. Audio Eng.
Soc. Vol. 36, No. 4, 1988
[10] Regulamento dos Requisitos Acústicos de Edifícios. Building Acoustics Noise
Code- in Portuguese., Law nº 129/2002 of 11 de May 2002

You might also like