You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/343224746

COMPARING METHODS FOR DESCRIBING LOW- FREQUENCY AIRBORNE


ISOLATION IN ISO AND ASTM STANDARDS

Conference Paper · July 2019

CITATIONS READS

2 148

2 authors:

John Loverde Wayland Dong


Veneklasen Associates, Inc. Veneklasen Associates
105 PUBLICATIONS 98 CITATIONS 120 PUBLICATIONS 110 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by John Loverde on 26 July 2020.

The user has requested enhancement of the downloaded file.


COMPARING METHODS FOR DESCRIBING LOW-
FREQUENCY AIRBORNE ISOLATION IN ISO AND ASTM
STANDARDS
John LoVerde and Wayland Dong
Veneklasen Associates, Santa Monica, California USA
email: wdong@veneklasen.com
The most common descriptor of airborne noise insulation in North America is sound transmission class (STC)
defined in ASTM E90 and E413. In most of the rest of the world, the common index is weighted sound reduction
index (Rw) defined in ISO 717-1. ISO 717-1 defines two spectrum adaptation terms which are intended to better
account for different source spectra, for example, traffic noise with elevated levels at low-frequencies. These
various ratings and methods are compared with a large set of sound isolation test data. Given these modified
ratings systems, there appears to be a need to be more specific regarding airborne noise isolation, especially at low
frequencies. This paper will determine if modifications to the STC rating may improve the information being
provided to engineers to evaluate acoustical performance.
Keywords: airborne noise isolation, low-frequency, single number rating

1. Introduction
The single-number rating used to describe airborne noise isolation in North America is sound trans-
mission class (STC), defined in ASTM E413.1 Although it is widely used, there remain some questions
regarding its suitability when it comes to isolation at the lower frequencies. The equivalent rating used
in most of the rest of the world is the weighted sound reduction index (Rw), defined in ISO 717-1.2 The
ratings are very similar but there are some significant differences. Additionally, the ISO standard defines
spectrum adaptation terms which can be used to evaluate low-frequency airborne noise isolation. This
paper uses existing test data to examine the differences between these ratings. The long-term goal is to
determine if STC is providing a suitable method for evaluating airborne noise isolation, particularly the
low frequencies, or if changes are warranted. Considerable work remains to address this goal. In this
paper, we present a selection of results that came out of this analysis.

2. Comparison of STC and Rw


STC and Rw are very similar, but have some differences that can affect the rating with respect to low
frequencies.

2.1 Reference Curve


STC and Rw are calculated using a similar reference curve. Both are 16 third-octave bands wide, and
the single-number rating is determined by the level of the reference curve at 500 Hz. However, the ASTM
curve runs over the 125 to 4000 Hz bands, while the ISO reference curve runs over the 100 to 3150 Hz

1
ICSV26, Montreal, 7-11 July 2019

bands. The curves are identical in the 15 bands of overlap. Northwood states, “with an eye to international
standardization, it was decided to adopt the shape of the DIN curves as the basis for the ASTM sound
transmission class contours.”3 Despite this, the ASTM reference curve is in the 125-4000 Hz range, even
though the DIN reference spectrum had the same 100-3150 range as the current ISO standard.
It appears that the original ASTM reference curve was defined in half-octaves instead of third-octaves,
which might account for the difference. Although if that is the case, it is not clear why the frequency
range was not adjusted when the method was changed to third-octave bands. It is clear that in the original
studies that led to the STC rating that low-frequency noise was not a concern, so perhaps the difference
was not considered significant. The justifications for the rating system were based on the spectrum of
“standard household noise” described as diminishing at 4 to 6 dB per octave below 250 Hz.3
Today there are many sources, notably home theatre subwoofers, but also sound systems of all types
(televisions, stereos, portable Bluetooth loudspeakers, etc.), that are capable of generating significant
energy below 250 Hz. Additionally, there are double-leaf assemblies in which the resonance of the air-
space falls in the range of the 100 or 125 Hz bands, so the lower frequency limit has the potential to cause
significant differences in rating for some assembly types.

2.2 Curve-fitting Method


The second difference is in the curve fitting method. In both standards, the measured transmission
loss is plotted against the reference curve, and the difference of the measured spectrum below the refer-
ence curve are the deficiencies at that band. In both standards, the reference curve is raised in 1 dB steps
until the sum of deficiencies does not exceed 32, or 2 dB per band. The ASTM reference also includes a
criterion that the deficiencies do not exceed 8 in any band.
According to an analysis by Clark, the 8-dB rule is “a vestigial remnant of an earlier version” of the
rating, and is “unnecessary.”4 Sepmeyer writes that the 8-dB rule is “obviously overrestrictive” and that
the 2 dB per band requirement is “adequate except for dips of 10 dB or more near 2 kHz.”5 The studies
in Ref. 4 and 5 were based on listening tests. Although based only on anecdotal and informal listening
while testing, the authors’ experience is also that dips in the transmission loss spectrum at single frequen-
cies do not make a large subjective difference. We are in agreement that the 8-dB rule is overly conserva-
tive and should be removed from the rating.

2.3 Comparison of Ratings


Data was compiled for over 7000 transmission loss tests performed at Western Electro-Acoustic La-
boratory (WEAL), which is affiliated with the authors’ company. The test specimens included stud walls,
doors, glass, and window assemblies. The calculated STC and Rw ratings are shown in Fig. 1 both with
the existing STC rating (a) and with the 8-dB removed (b).

2 ICSV26, Montreal, 7-11 July 2019


ICSV26, Montreal, 7-11 July 2019

80 80
a) b)
70 70

60 60

STC (no 8 dB rule)


50 50
STC

40 40

30 30

20 20

10
10

0
0
0 20 40 60 80
0 20 40 60 80
Rw Rw

Figure 1. a) Relationship between STC and Rw for all assemblies. b) with the STC rating calculated without the
8 dB rule. The rule is responsible for the majority of the difference between the ratings.
0.800

0.700

0.600

0.500

0.400

0.300

0.200

0.100

0.000
-5 -4 -3 -2 -1 0 1 2 3 4 5

STC-Rw STCno8-Rw

Figure 2. Histogram of the difference between STC and Rw ratings, with STC calculated as written and without
the 8-dB rule.

2.4 Comparison with Other Laboratory


A similar analysis was performed by Höller based on test data collected at the National Research
Council, Canada (NRC).6 In broad terms the data are similar, and Figure 2 in Ref 6 looks very similar to
Figure 1 above. The statistics are shown in Table 1 and are similar between labs. However, there is a
noticeable bias between the data sets of the two laboratories. In both sets of data, there are a number of
tests where the STC is significantly lower than the Rw due to the 8-dB rule, resulting in a long tail on the
left side of the histogram in Figure 2. When this is removed by computing STC without the 8-dB rule,
the NRC data set has a significant bias towards tests where the STC was higher than the Rw, as the mean
difference is 1.1 dB. The WEAL data set has a mean very close to zero.

ICSV26, Montreal, 7-11 July 2019 3


ICSV26, Montreal, 7-11 July 2019

The reason for this difference is not apparent. It is conceivable that there is a systematic difference
between the laboratories that results in a bias in the 100 Hz band. For example, if the same specimen
measured in the two labs differs at 100 Hz, the labs could report the same STC rating but a different Rw
rating. However, this cannot account for the size of the difference, as it would require a change in the
value of the WEAL tests at 100 Hz of about 15 dB in order to replicate the bias reported in the NRC
tests. It could be related to the biases within the data sets of the two labs, i.e., different percentages of
tests of different assembly types. Both reported data sets are “all data” type sets without attempt to stand-
ardize to a typical or representative data set.
Table 1: Comparison of STC and Rw reported by two laboratories
WEAL NRC
Mean difference -0.14 0.23
Standard deviation 0.86 1.25
STC – Rw
R2 0.993 0.989
% within 1 point 93.5 ~80
Mean difference 0.11 1.1
Standard deviation 0.56 Not reported
STC (no 8-dB) – Rw
R2 0.998 Not reported
% within 1 point 97.8 ~80

3. Relationship between low-frequency bands and single-number rating


3.1 Variation within frequency bands
To investigate the behavior at low frequencies, the data set was analyzed to examine the distribution
at the individual frequency bands. Since STC 50 is the minimum code requirement in the USA, it was
natural to start with assemblies having this rating. Assemblies with STC ratings between 49-51 (the wider
range allowed for a larger data set and is consistent with repeatability variation of the method) were
selected (n=327). Figure 3 shows histograms of the TL for the STC-50 assemblies at selected frequencies.
(Note that the values are for third-octave bands; the bands not shown had the same pattern but are omitted
for clarity.)
Figure 3 shows that the variation in each band appears to be distributed approximately normally at
most frequencies. The mid-frequency bands have relatively narrow distributions, with the width increas-
ing towards the low and high frequencies. The standard deviation of the distribution at each frequency
band is graphed in Figure 4. Some of the increase at low frequencies can be attributed to increased un-
certainty in the measurement; however, this is not the only effect. These distributions show the range of
values that can still result in a rating of STC 50. It appears that the mid frequencies are the most important
in this regard, as the distributions here are narrow, while the low-frequency performance can vary more
without affecting the rating. This variation in low-frequency performance may provide motivation for
modifying the rating, or developing a new rating that accounts for performance at these frequencies.

4 ICSV26, Montreal, 7-11 July 2019


ICSV26, Montreal, 7-11 July 2019

120

100

80
Count

60

40

20

0
10 14 18 22 26 30 34 38 42 46 50 54 58 62 66 70 74 78
Transmission loss (dB)

63 80 100 125 160 200 500 1000 2000 4000

Figure 3: Histograms of the transmission loss at selected third-octave bands for the subset of assemblies with STC
ratings between 49-51.
7.0

6.0
Standard deviation (dB)

5.0

4.0

3.0

2.0

1.0

0.0
63

1600
80
100
125
160
200
250
315
400
500
630
800
1000
1250

2000
2500
3150
4000
5000

Figure 4: Standard deviation of the data in each third-octave band for the subset of assemblies with STC ratings
between 49-51.

3.2 Correlation between bands


Of course the bands below 125 Hz do not affect the STC rating, but they are also not completely
independent of the adjacent bands that do affect the rating. In fact, for most assemblies, the shape of the
transmission loss curve is relatively the same. To investigate this on a broad level, the coefficients of
determination (R2) were computed between the TL in each third-octave band and the STC rating. See
Figure 5 (blue curve). There is very high correlation between the TL of the mid-frequency bands and the
STC rating. This is to be expected, as the STC rating was originally studied for the purpose of measuring
isolation in these bands (see Ref. 3 and 4.)

ICSV26, Montreal, 7-11 July 2019 5


ICSV26, Montreal, 7-11 July 2019

1.0

Coefficient of determination (R2)


0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0

2500
63
80
100
125
160
200
250
315
400
500
630
800
1000
1250
1600
2000

3150
4000
5000
STC Adjacent band

Figure 5: Coefficients of determination (R2) between the TL in each third octave band with STC rating (blue),
the band immediately higher in frequency (red).
What is perhaps surprising is that the correlation with STC rating remains high even at the ends of the
reference curve (R2=0.72 at 125 Hz, for example). Also as expected, the value drops steeply below the
125 Hz lower limit of the reference curve. This can be explained by examining the correlation between
adjacent bands (red curve in Figure 5). The decrease in the blue curve from 125 to 100 Hz is determined
by the correlation between the TL in the 125 and 100 Hz bands, and so on.
The behavior can also be seen by plotting the TL at individual third-octave bands as shown in Figure
6 for some example bands. At 630 Hz, there is very strong correlation between the TL and the STC
rating. The standard deviation of the STC rating at a given TL (i.e., the “width” of the distribution in the
vertical direction in the figure) is only about 2.5 points at 630 Hz. At 125 Hz, the distribution is at least
twice as wide, and from the figure it is seen that the 8-dB rule enforces a linearity on the low (left hand)
side of the group. At 63 Hz there is no useful correlation.

6 ICSV26, Montreal, 7-11 July 2019


ICSV26, Montreal, 7-11 July 2019

80

60
STC

40

20
630 Hz
0
80

60
STC

40

20
125 Hz
0
80

60
STC

40

20
63 Hz
0
0 20 40 60 80
TL in band

Figure 6: Relationship between STC rating and TL in third-octave bands

REFERENCES
1. ASTM Standard E413, ‘Classification for Rating Sound Insulation’. (ASTM International, 2010).
2. ISO 717-1, Rating of sound insulation in buildings and of building elements -- Part I: Airborne Sound Isolation.
(International Standards Organization, 1996).
3. Northwood, T. D. Sound-Insulation Ratings and the New ASTM Sound-Transmission Class. J Acoust Soc Am
34, 493 (1962).
4. Clark, D. M. Subjective Study of the Sound‐Transmission Class System for Rating Building Partitions. J.
Acoust. Soc. Am. 47, 676–682 (1970).
5. Sepmeyer, L. W. Study of the sound transmission class system for rating building partitions—Another view. J.
Acoust. Soc. Am. 80, 1404–1407 (1986).
6. Hoeller, C. Review and Comparison of ASTM and ISO Standards on Sound Transmission in Buildings. in Proc
INTER-NOISE 2018 (2018).

ICSV26, Montreal, 7-11 July 2019 7

View publication stats

You might also like