Open plan office acoustics - a multidimensional optimization problem

Abstract

The international standard ISO 3382-3:2012 defines a number of measurable room acoustic parameters for the objective evaluation of the acoustics of open plan offices. The main acoustical problem is distraction by speech and conversation between other people. However, this is not a simple one-dimensional noise problem that can be solved by a sufficiently high damping of the room. If the reverberation time is very short, the remote voices are heard with high clarity and thus the amount of distraction is high. But a long reverberation time leads to a very noisy environment, which is also disturbing. Similarly with the background noise: It should be neither too low nor too high. The most interesting acoustic parameters are the distraction distance and the privacy distance, both derived from the Speech Transmission Index (STI). These parameters depend on the important acoustic parameters, namely the amount of absorption, the effect of screens, the spatial attenuation and the masking from background noise. For the acoustical design, it is suggested to take the background noise from human activity into account, assuming a vocal activity that depends on the type of office.

Full text

DAGA 2018, 19-22 March 2018, Munich, Germany
Open plan office acoustics – a multidimensional optimization problem
Jens Holger Rindel
Odeon A/S, Scion-DTU, DK-2800 Kgs. Lyngby, Denmark, E-Mail:jhr@odeon.dk
Introduction
The international standard ISO 3382-3:2012 [1] defines a
number of measurable room acoustic parameters for the
objective evaluation of the acoustics of open plan offices.
Four quantities are mandatory:
• distraction distance r
D
• spatial decay rate of A-weighted sound pressure
level (SPL) of speech, D
2,S
• A-weighted SPL of speech at 4 m, L
p,A,S,4 m
• average A-weighted background noise level, L
p,A,B
.
The distraction distance is defined as the distance from a
source within which the speech transmission index (STI) ≥
0.5. In addition, two quantities may be determined; STI in
nearest work station, and the privacy distance, r
P
(STI = 0.2).
The main acoustical problem is distraction by speech and
conversation between other people. However, this is not a
simple one-dimensional noise problem that can be solved by
a sufficiently high damping of the room. If the reverberation
time is very short, the remote voices are heard with high
clarity and thus the amount of distraction is high. However, a
long reverberation time leads to a very noisy environment,
which is also disturbing. Experiments in the laboratory have
shown that excessive sound absorption may have a negative
impact on occupants’ perception of noise, acceptability of
noise and performance of office work [2].
Similarly with the background noise: It should be neither too
low nor too high. People may be concerned by silence as
much as with the noise [3].
The distraction distance is a particularly interesting parameter
because it takes into account most of the important acoustic
parameters, namely the amount of absorption, the effect of
screens, the spatial attenuation and the masking from
background noise. The background noise in the empty office
must be applied for measurements in accordance with ISO
3382-3, but the standard opens for additional calculation of
results using other kinds of background noise, e.g. that from
human activity in the office.
Noise from human activities
The French standard NF S 31-199 [4] divides open-plan
offices into four categories with different type of activity and
thus with different levels of noise from human activity:
• mainly telephone, L
A,eq
= 48 dB to 52 dB,
• collaborative work, L
A,eq
= 45 dB to 50 dB,
• individual work, L
A,eq
= 40 dB to 45 dB,
• receiving the public, L
A,eq
< 55 dB.
Measured noise level and time distribution for various office
activities were reported together with a prediction model for
human activity noise in an office [5]. Talking was the source
with highest sound pressure level, L
p,A
= 58 dB in 1 m
distance. Another empirical model for human activity noise
was derived from measurements in five offices [6].
Disturbance by speech noise
A recent investigation in 21 Finnish offices [7] showed that
the percentage highly disturbed (% HD) by speech was
correlated with the distraction distance and other pameters,
see Table 1. The correlation with the background noise
(measured without human activity) was negative, i.e. higher
background noise causes less disturbance by speech. It is
remarkable that there was no correlation with the spatial decay
rate D
2,S
. The highest correlation was found with L
p,A,S,4 m
.
However, it was concluded in [7] that the distraction distance
is the most relevant acoustic parameter for prediction of
disturbance by noise in open-plan offices.
Table 1: Relation between % HD by speech and the ISO
3382-3 parameters. Data from Haapakangas et al. [7].
Acoustic parameter R
2
for % HD by speech
r
D
(m) 0.294
D
2,S
(dB) 0.007
L
p,A,S,4m
(dB) 0.327
L
p,A,B
(dB) 0.266
Simulations with background noise from
speech
In order to through some light on the complicated relationship
between the acoustical properties and parameters, a series of
simulations was made using the ODEON room acoustic
software. The basic room model was the same as used in a
previous study [8]. The ceiling was either highly absorbing or
highly reflecting, yielding reverberation time around 0.4 s and
1.2 s, respectively. The rooms were fully furnished and either
without screens or with sound absorbing screens. Three
different heights of the screens were used, namely 1.2 m, 1.5
m and 1.75 m. The reverberation times with screens was a
little lower than without scteens, approximately 0.4 s and 1.0
s, respectively.
The simulations were made using four different source
positions and four related lines of receiver positions. The
results of each simulation was the room acoustic parameters
averaged over the four lines of sound propagation in
accordance with ISO 3382-3.

The background noise was varied from 30 dB (A-weighted)
to 60 dB in steps of 5 dB. Since the purpose was to simulate
noise from human activities, i.e. mainly speech, the spectrum
applied for the background noise was a typical speech
spectrum, see Figure 1.
Figure 1: Speech spectrum applied for the simulations. The
value at 63 Hz is from [9], the other values are from [1].
Results of simulations
First, we look at the results from offices without screens. The
distraction distance r
D
and the privacy distance r
P
are shown
as functions of the level of background noise in Figure 2.
While the distraction distance is relevant at noise levels below
50 dB, the privacy distance is more relevant in case of higher
levels of the background noise. The distraction distance
cannot be determined in the higher noise levels because STI
< 0.5 everywhere, even in the closest positions.
Figure 2: Distraction distance and privacy distance as
functions of background noise level. Results from computer
simulations.
Increasing the reverberation time has the effect that STI
decreases, and thus it is expected that r
D
and r
P
decrease.
While this is true for r
D
, the result is the opposite for r
P
. The
explanation is that the longer reverberation time decreases
STI in the close positions, but does not influence STI in the
remote positions, where the signal-to-noise level is low. Thus,
the slope of the STI/distance curve is more shallow with long
reverberation time than with a short reverberation time, and
the point where STI = 0.2 moves to a longer distance.
For a good acoustical design the goal is a short distraction
distance, preferably below 5 m as recommended in Annex A
of ISO 3382-3. Although this might be obtained with a very
long reverberation time, this is clearly an unacceptable
solution.
Another possibility is a very high background noise level,
nearly 50 dB if the reverberation time is 0.4 s. This
corresponds to the first category of open-plan offices in NF S
31-199 [4] where the assumed noise from human activity is
between 48 dB and 52 dB.
Another way to approach the goal is to introduce noise screens
between the workstations. This reduces the distraction
distance, see Figure 3. The effect of the screens is most
pronounced in low background noise.
Figure 3: Relation between distraction distance and
background noise level. Full lines: Simulations without and
with screens of various heights. Dots: Measured data from
Haapakangas et al. [7].
Figure 3 also shows the results from the 21 Finnish offices
reported in [7]. In general, these measured results suggests
screens that are high and closer together than those applied in
the simulated offices. However, the relationship between
background noise and distraction distance in the simulations
has a slope, which is similar to the slope of the linear
regression line through the measured data. The slope is
approximately -1 dB per m (R
2
= 0.69).
The effect of screens is illustrated in Figure 4, which
combines the two parameters distraction distance and spatial
decay rate of speech. It is seen that increasing the
reverberation time (RT) leads to a decrease of both r
D
and
D
2,S
. Introduction of screens and increasing the screen height
leads to a decrease of r
D
and increase of D
2,S
. With high
screens, the RT has strong influence on D
2,S
, but rather limited
influence on r
D
.
26,6
32,5
36,9
40,6
34,6
27,4
21,4
16,1
0
10
20
30
40
50
63 125 250 500 1000 2000 4000 8000
SPL, dB
Octave band centre frequency, Hz
Lp,A = 40 dB
0
10
20
30
40
30 40 50 60
Distance (m)
Background noise level (dB)
Distraction distance (T = 1.2 s)
Privacy distance (T = 1.2 s)
Distraction distance (T = 0.4 s)
Privacy distance (T = 0.4 s)
25
30
35
40
45
50
0 5 10 15 20
Lp,A,B (dB)
Distraction distance, rD(m)
T = 0.4 s
T = 0.4 s, screens 1.2 m
T = 0.4 s, screens 1.5 m
T = 0.4 s, screens 1.75 m
Measured, 21 offices
Linear (Measured, 21 offices)

Figure 4: Relation between distraction distance and spatial
decay rate D
2,S
in simulations with different reverberation
times (RT), with and without screens. The background noise
level is 35 dB with speech spectrum.
Figure 5 is similar to Figure 4, but here the results are shown
for three different levels of the background noise. The
decrease of r
D
with increasing background noise is very
pronounced. In high background noise and with screens, RT
influences D
2,S
, only, not r
D
.
Figure 5: Relation between distraction distance and spatial
decay rate D
2,S
for different levels of background noise,
different reverberation times, with and without screens.
Figure 6 shows the results of the simulations in a diagram,
which combines the two parameters distraction distance and
SPL of speech at 4 m. Increasing screen height is efficient to
reduce r
D
, but has very limited influence on L
p,A,S,4m
.
Increasing RT has a strong influence on L
p,A,S,4m
, and the level
exceeds 50 dB if T > 0.4 s in this example. This diagram
explains why it is necessary to optimise the reverberation
time; it should neither bee too long nor too short.
Figure 6: Relation between distraction distance and SPL of
speech at 4 m in simulations with different reverberation
times, with and without screens. The background noise level
is 35 dB with speech spectrum.
Discussion
Annex A of ISO 3382-3 contains guidelines for evaluating the
acoustic parameters in offices. Conditions are good if r
D
≤ 5
m, D
2,S
≥ 7 dB and L
p,A,S,4 m
≤ 48 dB. Conditions are poor if
r
D
> 10 m, D
2,S
< 5 dB and L
p,A,S,4 m
> 50 dB.
Figure 7 is in a diagram that displays D
2,S
against r
D
. The data
from open-plan offices, in which the % HD by speech was
reported, Haapakangas et al. [7] are plottet. Most offices are
neither poor nor good according to the guidelines. D
2,S
> 6 dB
in most offices and the whole picture is scattered. The
relevance of the suggested limits is not obvious.
Figure 7: Relation between distraction distance and spatial
decay rate D
2,S
in measured data from open-plan offices,
Haapakangas et al. [7]. The colour of the dots indicate the %
HD by speech. The shaded areas refer to “good” (light green)
and “poor” (purple) in Annex A of ISO 3382-3 [1].
In Figure 8 the same data are displayed in a diagram with
L
p,A,S,4 m
and r
D
as the acoustic parameters. This diagram
makes more sense, because there is a tendency that the offices
with the worst subjective rating are in the upper left corner.
0
1
2
3
4
5
6
7
8
10 15 20 25 30
D2,S (dB)
Distraction distance, rD(m)
No screens 1.2 m screens
1.5 m screens 1.75 m screens
T = 0.4 s T = 1 s
0
1
2
3
4
5
6
7
8
0 5 10 15 20 25 30
D2,S (dB)
Distraction distance, rD(m)
Lp,A,B = 35 dB Lp,A,B = 40 dB
Lp,A,B = 45 dB T = 0.4 s
T = 1 s
Increase background noise
40
45
50
55
60
10 15 20 25 30
Lp,A,,S,4 m (dB)
Distraction distance, rD(m)
T = 1 s T = 0.4 s
No screens 1.2 m screens
1.5 m screens 1.75 m screens
0
2
4
6
8
10
12
14
0 5 10 15 20
D2,S (dB)
Distraction distance, rD(m)
0-20 % HD 20-40 % HD 40-60 % HD > 60 % HD

However, the correlation between the subjective rating and r
D
is not very obvious. The offices with best rating (0-20 % HD)
have r
D
between 2.5 m and 14 m. This can be the result of
very different background noise in the measured offices. As
remarked previously (Figure 3) there is a rather strong
correlation between and r
D
and background noise. Actually,
some of the best-rated offices had short r
D
, but also quite high
background noise.
Figure 8: Relation between distraction distance and SPL of
speech at 4 m L
p,A,S,4 m
in measured data from Haapakangas
et al. [7]. The colour of the dots indicate the % HD by speech.
The shaded areas refer to “good” (light green) and “poor”
(purple) in Annex A of ISO 3382-3 [1].
Figure 9: Relation between distraction distance adjusted to
40 dB background noise level and SPL of speech at 4 m
L
p,A,S,4 m
in measured data from Haapakangas et al. [7]. The
colour of the dots indicate the % HD by speech. The shaded
areas are suggested for “good” (light green) and “poor”
(purple).
To overcome this dependency of the background noise, it is
suggested to ‘normalise’ the distraction distance in such a
way, that it represents a realistic background noise from
human activity. Using the correlation from Figure 3, the
adjusted distraction distance at 40 dB background noise from
speech, r
D,40
can be roughly approximated by:
(
)
BA,,D40,D
40
p
Lrr −−≈
[m]
(1)
This leads to the diagram in Figure 9. It is suggested that the
corresponding limit for ‘good’ conditions is r
D,40
≤ 9 m.
Normally, Equation (1) should not be used. The STI results of
measurements as well as simulations should be calculated
directly using the 40 dB speech noise spectrum, see Figure 1.
Conclusion
Among the acoustic parameters, the spatial decay rate D
2,S
seems to have no relevance. However, the SPL at 4 m is an
important design parameter together with the distraction
distance. Since the latter is strongly dependent on the
background noise level, it is suggested to ‘normalise’ the
distraction distance to a well-defined level of background
noise that may better represent the noise from human
activities.
References
[1] ISO 3382-3 (2012). Acoustics – Measurement of room
acoustic parameters – Part 3: Open plan offices.
International Organization for Standardization, Geneva,
Switzerland.
[2] I. Balazova, G. Clausen, J.H. Rindel, T. Poulsen, D.P.
Wyon (2008). Open-plan office environments: A
laboratory experiment to examine the effect of office
noise and temperature on human perception, comfort and
office work performance. Proceedings of Indoor Air
2008, 17-22 August 2008, Copenhagen, Denmark.
[3] V. Acun & S. Yilmaez (2018). A grounded theory
approach to investigate the perceived soundscape of
open-plan offices. Applied Acoustics 131, 28-37.
[4] NF S 31-199 (2016). Acoustic performance of open-plan
offices. AFNOR, Paris, France.
[5] T. Vervoort & M. Vercammen (2015). Background noise
level to determine the speech privacy in open plan
offices. Proceedings of Euronoise 2015, Maastricht,
Netherlands, 1209-1214.
[6] T.S. Dehlbæk, C-H Jeong, J. Brunskog, C.M. Petersen,
P. Marie (2016). The effect of human activity noise on
the acoustic quality in open plan office. Proceedings of
Internoise 2016, Hamburg, Germany, 3517-3526.
[7] V. Haapakangas, V. Hongisto, M. Eerola, T. Kuusisto
(2017). Distraction distance and perceived disturbance
by noise – An analysis of 21 open-plan offices. J.
Acoust. Soc. Am. 141, 127-136.
[8] J.H. Rindel, C.L. Christensen (2012). Acoustical
simulation of open-plan offices according to ISO 3382-
3. Proceedings of Euronoise 2012, Prague, Czech
Republic.
[9] J.H. Rindel, C.L. Christensen, A.C. Gade (2012).
Dynamic sound source for simulating the Lombard
effect in room acoustic modeling software. Proceedings
of Internoise 2012, New York, USA.
40
45
50
55
0 5 10 15 20
Lp,A,S,4 m (dB)
Distraction distance, rD(m)
0-20 % HD 20-40 % HD 40-60 & HD > 60 % HD
40
45
50
55
0 5 10 15 20
Lp,A,S,4 m (dB)
Distraction distance at 40 dB, rD,40 (m)
0-20 % HD 20-40 % HD 40-60 & HD > 60 % HD