December_1970 — Page 19

provided certain specialised facilities are available. At low frequencies most sources will be omnidirectional (i.e. radiate equally in all directions) but at high frequencies most sources will begin to radiate more in one direction than another. Figs. 3a and 3b show the directional properties of an electric typewriter and the human voice. Dia- grams such as these can tell us in what circumstances it may be possible to 'point' a noise source away from a noise sensitive area.

Multiple sources, such as typing pools, may be 'added up' in the usual way: N sources each producing a level of LdB will produce together a total level of (L+ 10 log N).

Sound transmission

Transmission of sound in the land- scaped office presents a totally dif- ferent picture from that in a conven- tional office. In a normally propor- tioned room with large non-absorbing surfaces, sound from the source is repeatedly reflected from the walls and ceiling to form a reverberant field throughout the room, of roughly con- stant loudness.

Direct sound only predominates close to the source; increasing the amount of absorption in the room will lower the reverberant level, but cannot affect the direct sound. This is illus-

trated in Fig. 4a. The reverberant sound is transmitted through parti- tions into adjoining offices, where a second reverberant field is set up, but at a level determined by the transmis- sion loss of the partition.

This kind of behaviour cannot be expected in landscaped offices because of their unconventional shape. We have therefore been carrying out re- search at the University of Salford to discover how sound does behave in such offices, by measurements in both full size offices and in models, and by theoretical calculations on the KDF 9 computer. The experiments and cal culations have all shown the same varia- tion of loudness with distance from the noise source, though the precise shape of the curve depends upon the frequency of measurement, the nature of the ceiling and floor finishes, and on the floor to ceiling height. We can summarise the common features of these investigations as follows (see Fig. 4b): (1) Near Field (D): out to 1 or 2m from the source the direct sound predominates, so that there is a 6dB fall for each doubling of distance from the source. (2) Plateau Region (P): between about 1 or 2m and 8 or 10m from the source the level sometimes falls more slowly, at about 3dB for distance doubling. (3) Far Field (F): between about 8 or 10m and 20 or

30m from the source the level again falls more rapidly, again approaching 6dB for distance doubling.

Beyond 20 or 30m from the source, or sooner in small offices, the sound level ceases to fall, due to reflec- tion from the walls setting up a pseu- do-reverberant field. In extremely large offices the level continues to fall, but at a decreased rate due to the in- creasing reflectivity of the surfaces at grazing incidence.

Although the relative size of these features varies with frequency, and from office to office, they have been apparent to a greater or lesser degree in all our investigations to date. Typically, Fig. 5 shows a complete scan of the sound field in a tenth scale model office, with a noise source at the centre. The drawing is isometric, with vertical height representing sound level, and the major regions discussed above can be identified.

As might be expected, smaller office dimensions tend to take condi- tions back towards the reverberant room situation. Measurements in the model with decreasing width showed an increase in level at points in the far field, but little change on the ‘plateau'. The limit seems to be reached when the width falls below about 12m. Al- teration of the ceiling height affects the loudness level of the 'plateau', and

T.L.

(a) CONVENTIONAL

Far East BUILDER, December 1970

Partition

R.R.

INTENSITY LEVEL

Av.

P.

Av.

Av.

F

(b) LANDSCAPED

Fig. 4a & b. Sound field in conventional and landscaped offices

19