Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Journal of Sound and Vibration (1980) 69(4), 603-610

ASSESSMENT OF NOISE ANNOYANCE:


THE INTRODUCTION OF A THRESHOLD LEVEL IN
L,, CALCULATIONS

T. GJESTLAND AND G. OFTEDAL

Acoustics Laboratory, ELAB, The University of Trondheim, Norwa)

(Received 17 September 1979, and in revised form 3 November 1979)

A modification of L,, is suggested. The modification is based on the simple fact that noise
below a certain threshold may not be heard, and consequently cannot contribute to the
annoyance. Results from laboratory experiments show a reasonably good correlation with
subjective evaluation. The proposed index may give a new basis for the understanding of
the effects of noise control measures.

1. INTRODUCTION
The energy equivalent continuous A-weighted sound level is the noise description most
commonly used in connection with traffic noise and other types of community noise [l].
Several investigations on community reaction to noise, however, indicate that L,, does not
correlate with subjective assessments in all practical noise situations. This is especially the
case when the frequency or time distribution of the noise is somewhat “unusual”.
Comparisons of different noise situations with equal L,, show significant preferences one
way or another [2]. This leads to the conclusion that other parameters besides L,, should
be taken into account. Additional descriptors, and corrections or limitations in the use of
L,, have been proposed [3,4].
A simple modification of L,, is shown to be a possible solution to the above mentioned
problem [5].

2. EQUIVALENT NOISE LEVEL ABOVE A THRESHOLD


The regular equivalent noise level is calculated from the equation

L,, = lolg&~filoL~‘l”. (1)


I
Here fi represents the fraction of time (in percent) the noise level equals L,.
The instantaneous noise level will normally vary quite a bit, typically 20-30 dB(A) for
road traffic noise. Part of the time the noise may therefore be below a masking threshold
originated by the presence of other “non-annoying” sound sources, or simply below the
threshold of hearing.
Only audible noise may contribute to the annoyance. One may thus introduce a specific
noise threshold level, L-I; and when assessing the annoyance take into account only those
periods when the actual noise signal exceeds this threshold level. A modified equivalent
noise level above a threshold, LTeq, can be calculated simply by omitting the low level
603
0022-460X/80/080603 t08 $02.00/O @ 1980 Academic Press Inc. (London) Limited
604 T. GJESTLAND AND G. OFTEDAL

periods:
L Teq= lolg~~filo=~“O, LiceCo for LiCLT. (2)
I
Figures 1 and 2 show the “active” periods with a specific noise signal and two different
threshold levels. Figure 3 shows L ~~~as a function of LT. For low threshold levels LTeq
approaches the regular L,,. This corresponds to a situation where the noise is audible all
the time. For increasing threshold levels LTeqdecreases and drops infinitely for LT equal to
and above the maximum noise level, corresponding to a situation where the noise is not
heard at all.

Figure 1. Noise level versus time. Shaded areas indicate periods when the noise level exceeds the threshold
level LT1.

t-

Figure 2. Noise level versus time. The threshold level L r2 has been reduced compared to Lrr and the “active”
periods increase accordingly.

Threshold level Lr (dB(A))

Figure 3. L, as a function of threshold level. For Lr s LrminLreq equals L,,. The two different situations
indicated by 1 and 2 refer to Figures 1 and 2 respectively.
ASSESSMENT OF NOISE ANNOYANCE 60.5

3. JUSTIFICATION OF THE THRESHOLD CONCEPT


There is always a lower limiting threshold level below which a specific signal cannot be
heard. In a completely silent environment this threshold is equal to the regular hearing
threshold as established by audiometry. Under more “normal” conditions, however, a
number of different sound sources may be present, the sound from which is usually not
considered annoying or disturbing. Typical examples of such sources are low level
household appliances, radio and TV sets, speech, noise from general activity, etc. These
sound levels are generally rather high and, in most practical instances, these indoor sound
sources will partly mask indoor levels from outdoor sources [6]. Practical experience also
suggests that there is a certain reaction threshold or acceptance threshold and that noise
below this threshold, although detected, is not considered to be contributory to the
annoyance.
A simple recording of outdoor L,, for assessing the annoying effect of, say, traffic noise
indoors is thus basically wrong, as only the audible parts of the noise can contribute to the
annoyance. On this basis, one can justify the introduction of a threshold level, LT, and the
omission of the periods when the noise level from the annoying source is below this level.
Lr is a rather subjective parameter and may vary quite a bit. During rest periods and the
process of falling to sleep, LT must be very low, say in the range 25-35 dB(A) [l]. For
general daytime activities like conversation, working with household appliances, listening
to background music, etc., LT may be increased accordingly, say to 40-50 dB(A). For high
sound level activities like children playing, listening to stereo music, etc., LT may be
increased further by at least 10 dB(A) [6].
The “positive” background sound level is usually a time varying signal just like that from
the annoying source. The threshold level must be sufficiently low to allow for more quiet
periods of the masking signal, and will probably have to be below the average level of the
masking sound depending on the situation. LT is thus a function of “place and activity”
and depends among other things upon the information content of the masking sound
signal.
At first this may sound complicated but it should not be too difficult to establish a few
values of LT that may be applied when evaluating the annoyance for different situations.

4. DIFFERENT NOISE SITUATIONS WITH EQUAL L,,


In connection with road traffic noise the same equivalent noise level may be obtained for
different combinations of traffic volume and observation distance. In Figure 4 LTeq is
plotted for four different situations. The values were calculated by using equation (2) and
the source level of one car passing at a speed of 50 km/h. The source level of a truck is
considered to be 10 dB(A) above that of a light vehicle. The regular L,, is always the same,
but the critical threshold level above which the noise is totally masked varies more than
25 dB(A).
One may choose LT =25 dB(A), 40dB(A) and SOdB(A) for rest, quiet and busy
daytime activities respectively. Traffic noise levels are normally referred to as outdoor
levels. One can therefore transform threshold levels to an outdoor situation by adding, say,
15 dB(A) corresponding to the transmission loss from outdoors to indoors. (The same
effect could be obtained by reducing the traffic noise levels accordingly.) By using Figure 4
one can now evaluate the annoyance in the different situations.
For rest periods, LT = 25 + 15 = 40 dB(A), all four traffic situations seem equally
annoying with LTeq = 52 dB(A). For quiet daytime activities, Lr = 55 dB(A), the noise
from situation no. 1 is not heard, and consequently the noise cannot be annoying at all. The
noise from situation no. 2 is partly masked, LTeq = 50 dB(A), and this situation seems
606 T. GJESTLAND AND G. OFTEDAL

40 50
ThresholdlevelLr (dB(A))

Figure 4. .L, as a function of threshold level for four different traffic noise situations with equal L,,. See text
for further explanation. 0, 50 m 1000 cars/hour; 0, 25 m, 300 cars/hour; 0, 12 m, 90 cars/hour; @,, 9 m, 9
trucks/hour.

somewhat less annoying than the last two situations, where one still observes LTeq,=
52 dB(A). For busy daytime activities, L T = 6.5 dB(A), the noise from traffic situations no.
1 and 2 are completely masked and the noise from no. 3 is partly masked, the resulting LTeq
being reduced to 40 dB(A). The corresponding value for situation no. 4 is almost
unchanged, 51 dB(A).
This example clearly indicates that the regular L,,hardly gives a complete description of
a noise situation, and, in contrast, demonstrates the usefulness of the new concept, LTeq.

5. EXPERIMENTAL VALIDATION

A number of experiments have been reported in which attempts have been made to
correlate the equivalent level for different noise situations with a subjective assessment of
the annoyance. Most experiments include different noise situations with different Leq, and
in those cases L,, seems to correlate reasonably well [7]. However, when using different
noise situations with equal L,,the results seem to indicate that there are other important
parameters that influence the annoyance [2].
Accordingly, an experiment identical to that reported in reference [7] has been carried
out, copies of the same test tapes having been used. However, the different traffic noise
recordings were presented at equal L,,levels.
Ten test subjects were used, divided into two teams. The subjects were all university
students with normal hearing between 20 and 30 years of age. The test subjects were
exposed to a certain noise for 30 minutes and at the end of each session they were asked to
evaluate the annoyance by completing a questionnaire. The evaluation was done on a 10
point scale covering the range “not annoyed” to “extremely annoyed”. During the
exposure, the test subjects were free to read or write but they were not allowed to
communicate. The presentation was repeated for four different values of L,,and two
different, artificially produced, threshold levels. Figure 5(a)-(d) show LTeqas a function of
LT for the different presentation levels.
Five different traffic noise recordings were used. Table 1 gives a description of the traffic
volume and measuring conditions.
During the experiment the noise levels were adjusted to give the same L,,-value for the
30 minutes period for all five situations (A-E). The presentation was repeated for four
different values of Leq:56, 50, 44 and 38 dB(A), and two different, artificially produced
threshold levels LT.For practical reasons the experiment was limited to 25 different
ASSESSMENT OF NOISE ANNOYANCE 607

zz
Q
m 50 60 iU 80 40 50 60 70 80
D
50

40

30

20

IO
20 30 40 50 60

L, (dB(A))

Figure 5. LTeq as a function of LT for the different traffic situations used in the laboratory experiment.
Presentation level: (a) 56 dB(A); (b) 50 dB(A); (c) 44 dB(A): (d) 38 dB(A). Dashed lines indicate threshold levels
used in the experiment.

TABLE I

Description of traffic situations used in the experiment; for more detailed information,
see reference [7]

Noise Road Recording Vehicles % Recorded


number We distance (m) per hour heavy L<,, (dB(A))

A Expressway 15 2625 17 73
B Rural road 15 912 4 69
C Highway 25 909 1.5 61
D Highway 25 179 17 58
E Expressway 250 3527 7 51

combinations altogether. All noise levels are referred to the actual condition in the
listening room.
The artificial threshold levels were produced by a steady state background noise signal
with a weighting curve shown in Figure 6. This noise was always present in the test room,
even during rest periods. Evaluation of this noise alone showed that it did not contribute to
the annoyance (mean value 1.3 on the annoyance scale for the highest level). At low traffic
noise levels, however, this background noise would to some extent mask the “annoying”
noise source, and consequently reduce the subjective annoyance.

6. BACKGROUND THRESHOLD LEVELS


The subjective threshold levels do not necessarily coincide with the objectively
measured noise levels. The corresponding level LT was therefore fitted to the data by
comparing the difference in annoyance for masked and unmasked noise presentation. In
608 T. GJESTLAND AND G. OFTEDAL

Figure 6. Relative band levels of background noise in the test room.

the unmasked case, the five different traffic noise samples were evaluated as almost similar
whereas in the masked situations there were significant differences.t
Two different masking levels were used corresponding to subjective thresholds of
approximately 43 dB(A) and 49 dB(A) respectively. These levels are indicated by dashed
lines in Figures 5(a)-(d). The actual background noise levels in the listening room were
38 dB(A) and 44 dB(A) respectively.

7. LTeq VERSUS ANNOYANCE


For each test situation, the corresponding LTeqwas found from Figures 5(a)-(d),
combining presentation level, threshold level and noise recording. As an example one sees
from Figure 5(c) that a presentation of traffic noise E at L,,= 44 dB(A) and a threshold
level of 49 dB(A) corresponds to LTeq= 20 dB(A).
These values for LT=~are plotted in Figure 7 along with the mean scores on the
subjective annoyance scale. The solid line indicates the connection between the equivalent

0 IO 20 30 40 50 G

Lr,# (da(A))

Figure 7. Results from the laboratory experiment. Subjective assessment of annoyance (usually a 10 point
scale) versus LTlq.

t The differences were checked by applying the Wilcoxon signed-rank test. In the masked situation, traffic
noise E, which was used to establish the threshold level, was different from all the others with a significance
probability ranging from 0.85 to’0.98.
ASSESSMENT OF NOISE ANNOYANCE 609

level above a threshold, LT=~, and the subjectively assessed annoyance on a 10 point scale.
This line represents a best fit curve to the data points.

8. PRACTICAL CONSEQUENCES
People concerned with noise control, especially reduction of traffic noise, have often
been intrigued by the apparently great subjective benefits from even marginal noise
reduction measures. This situation may be explained by using the proposed threshold
theory.
In connection with indoor noise annoyance due to outdoor road traffic, the actual noise
levels are usually rather low compared with the “non-annoying” household noise sources.
Such situations are described by the decreasing sections of the Lr-LT,, plots in Figures
5(a)-(d). Even a slight reduction in the noise level in the order of a few decibels,
corresponding to a similar small increase in the threshold level, may lead to a dramatic
reduction of LTeq and of subjective annoyance depending on the actual situation. A more
detailed explanation of these effects is given in reference [8].
In the United States the concept L DN has been adopted to compensate for the low
acceptability levels of noise during sleep periods. L DN gives a 10 dB penalty on night time
noise levels. The argument for doing this corresponds to that of lowering the night time
threshold level 10 dB below the daytime LT. The effect, however, may be rather different.
The LDN method will always increase Leq, whereas LTeq will increase only if the night time
noise levels exceed LT (i.e., can be heard). Low steady noise levels, like those from distant
traffic or other noise sources may still be below the threshold and should therefore not
contribute to an increase in the annoyance.

9. CONCLUSIONS
Most people seem to agree that the annoying effects of noise are a function of (at least)
noise levels and exposure time. Short periods at high levels seem to be more annoying than
relatively longer periods at lower levels.
The regular equivalent noise level, Leq, stresses the influence of high level periods and
seems to correlate reasonably well with subjective assessments of noise annoyance for
normal noise situations. However, L,, seems to fail in describing irregular situations,
especially at moderate and low noise levels.
It has been demonstrated here that the equivalent sound level as a noise descriptor with
respect to annoyance may be improved by introducing certain threshold levels. The
resulting modified noise index, L Teq, possesses the same properties as the regular L,, but
also takes into account whether or not the noise is audible to the exposed people. The
proposed modification is of importance for understanding the effects of noise control
measures.

ACKNOWLEDGMENT
This project was financed by the Royal Norwegian Council for Scientific and Industrial
Research.

REFERENCES
1. T. J. SCHULTZ 1978 Journal of the Acoustical Society of America 64, 377-405. Synthesis of
social surveys on noise annoyance.
2. G. W. CERMAK 1979 Journal of the Acoustical Society of America 65, 112-123. Exploratory
laboratory studies of the relative aversiveness of traffic sounds.
610 T. GJESTLANDAND G. OFTEDAL

3. D. W. ROBINSON1971 Journal of Sound and Vibration 14,279-298. Towards a unified system


of noise assessment.
4. K. MATSCHAT, E.-A. MULLER and G. ZIMMERMANN1977 Acustica 37, 267-272. On the
formulation of noise indices.
5. T. GJESTLAND1978 3rd International Congress on Noise as a Public Health Problem, Freiburg,
GBR, September 1978. Noise annoyance-shortcomings of L,,.
6. L. C. SUTHERLAND1978 Noise Control Engineering 11, 124-137. Indoor noise environments
due to outdoor noise sources.
7. T. JACOBSEN 1978 The Acoustic Laboratory, Technical University of Denmark, Report 24.
Measurement and assessment of annoyance of fluctuating noise.
8. T. GJESTLAND 1978 Electronic Research Lab. NTH, Norway, Report STF44A78030. Noise
annoyance.

You might also like