-85
-80
-70
ન
INTENSITY LEVEL
-65.
60
POSITION OF SOURCE
Rapid fall in level at first
Plateau
Second sharp falt
·BOUNDARY OF OFFICE
Gentle fall in far field
20
BOUNDARY OF OFFICE
the computer predicts a fall of about 2dB when the ceiling is raised from 3m to 4m. This is offset to some extent by a rise in the reverberant (far) field level.
Neglecting this finer detail of the sound field, a good engineering ap- proximation is to take an average rate of attenuation ('Av.' in Fig. 4b) of 4.5dB for distance doubling, beyond Im from the source. This has often been observed in practice, with an accuracy at any point of 1 or 2dB, and applies more or less independent of frequency, within this margin of error.
Thus if a level of XdB exists at 1m from the source, levels of (X-4.5), (X-9), (X−13.5) and (X–18)dB would be observed at 2, 4, 8, and 16m respec- tively.
Apart from propagation across the unobstructed office, we are also in- terested in the effects of barriers such as furniture and low screens. There is a fundamental limitation on the per- formance of a screen because of the effects of diffraction around it, and it is unlikely that even this limited per- formance will be achieved by low screens, owing to the effect of re- flections from the ceiling and floor.
Measurements we have made on 1.2m-high screens, both in models and in real offices, indicate that little at- tenuation can be expected for points further than 4m beyond the screen. But further research is required into this aspect of transmission problems.
Recipient response
The third aspect of the design problem concerns the response of re- cipients to noise. The major effects of noise are concerned with distraction/ arousal, annoyance, privacy and com- munication. These effects have long been studied in normal office environ- been studied in normal office environ- ments and the results of this work can probably be extended to landscaped probably be extended to landscaped offices, though such extension needs to be justified by further research.
The results for conventional offices have been expressed in terms of the
901
80
70
5
50
OCTAVE BAND INTENSITY LEVEL, dB
w
NC70
NC60
Fig. 5. Typical isometric sound plot of model of!
>
Noise Criteria (N.C.) curves. To use these, the octave band spectrum of noise at a point in the office has to be plotted on to special charts (see Fig. 6). The value of the highest N.C. curve crossed by the spectrum is then taken as the N.C. number of the noise. It is generally reckoned that the noise in a private office should not exceed N.C. 40, while that in general clerical of- fices should not exceed N.C. 45, if annoyance, distraction and inter- ference with telephone conversations are to be avoided.
However, the shape of the curves has been established by considering also the noise required to provide pri. vacy through masking; ideally, the noise in the office should have a spec- trum which lies just under the N.C. curve; low frequency noise will then be loud enough to mask overheard speech, without the high frequency noise becoming annoying.
In addition to continuous noise, particular care needs to be taken with NC50 noise sources such as telephone bells which have some 'alarm' content. The use of muted bells or flashing lights is generally recommended.
NC45
40
NC40
20
10
NC30
Discussion
When we apply the information NC20 given above to a design, two major difficulties arise. In the first place there is the straightforward problem of achieving sufficient attenuation of noise. Between conventional offices,
125 250 500 1000 2000 4000 8000
OCTAVE FREQUENCY, Hz
Fig. 6. N.C. curves, showing typical noise of N.C. value 50
Far East BUILDER, December 1970
Page 20Page 21
transmission loss varying from 15dB at low frequencies to 35dB at high fre- quencies is typical. But a transmission loss of 15dB will only be achieved be- tween two areas of a landscaped office if they are over 10m apart, the loss being more or less independent of frequency.
It is theoretically possible to im- prove on the figure of 4.5dB for dis- tance doubling, given particular ceiling properties, but ceilings with these properties have yet to be practically realised. Screens as used at present are ineffectual against a noise source more than 3m away: greater height would improve their performance, but might lead to increased levels by reflection from the screen.
The second problem involves the frequency independence of transmis sion loss in landscaped offices. When noise is transmitted through a parti- tion its spectrum is altered, the high frequencies being attenuated more strongly than the low frequencies. Consequently the noise spectra of common sources such as typewriters and speech are automatically adjusted to give an approximate fit beneath the N.C. curves.
80
870
OCTAVE BAND INTENSITY LEVEL, DB
E
70
50
30
10
T.L