Tracing a Tonal Noise Source on a Large Industrial Site. Paul Bassett MIOA.

Introduction 'We know that we have a tonal noise emission but we can't find out where it's coming from', explained the Environmental Manager at a large industrial processing factory, 'we need some help'. The presence of the tonal noise had been mentioned informally to the factory by an employee who lived nearby. Having confirmed that there was indeed a tonal noise, the company, in accordance with its commitment to the Chemical Industry Association's Responsible Care programme, was eager to trace the source of the tonal noise and undertake appropriate measures to reduce it. The Site The factory is located on the bank of the Dee Estuary in North Wales. Inland there is a housing estate in an elevated position which overlooks the industrial site and the estuary beyond. The factory, which is used for the manufacture of additives to soap powders and detergents, operates 24 hours a day on seven days a week. Between the factory and the housing is a busy coast road and railway line, but at night the background noise climate is lower and it was found that a tone was clearly discernible. During an initial site visit we were encouraged to find that the tone was a mid frequency ?hum? which tends to be easier to trace than a low frequency 'drone'. However, whilst the frequency of a tonal noise can often be readily identified at a considerable distance away from the noise source, actually measuring and tracing the source of the noise on an industrial complex can prove difficult due to the masking effect of a myriad other plant noise sources nearby. This was the case here. When one walked around the processing plant area at the factory, it was not possible to hear the tone. It was clear that, in order to trace the source of the noise, a tone matching technique using a narrow band frequency analysis would be required. The Approach The Environmental Manager feared that the noise source was likely to be one of the many chimney stacks on the site and that we would need to measure noise in the flues or on top of the stacks. Not being overly keen at this vertiginous project, we suggested that a more systematic approach should be adopted on terra firma. We considered that it should be possible to 'home in' on the area of the tonal noise by carrying out narrow band frequency analysis initially at ground level. Once the general area of the tonal noise had been defined it was hoped that further investigation at higher elevations on the plant would reveal the source of the noise. Failing this the contingency plan was to carry out a noise survey after the annual shutdown by assessing the noise as each plant process was switched on in turn. It was decided, therefore, to determine the frequency of the tone at the housing site by narrow band analysis, and then to repeat the narrow band analysis on the industrial site. The latter could have been done on a grid basis, but the company were also interested in noise levels at a number of specified plant areas around the site. Therefore it was agreed that the survey would comprise noise measurements and analysis at these points which were distributed over the whole industrial site. Noise Monitoring A noise monitoring point was established on the edge of the housing area with a clear line of sight to the factory. The survey was carried out late at night in order to minimise the influence of non-factory noise sources (road traffic, trains, etc). Noise levels were monitored using a Bruel & Kjaer Type 2231 Sound Level Meter connected to a Diagnostics Instruments PL21 Real Time FFT Analyser and Diconix printer. As well as ?A? weighted noise measurements this set-up enabled narrow band frequency analysis to be carried out in order to identify any problematic tonal elements. The noise from the factory site was characterised by a constant, and fairly broadband, background noise over which was superimposed a continual intrusive mid-frequency tonal noise which we estimated by ear to be around 500 Hz. The narrow band frequency analysis enabled this tonal element to be identified at a frequency of 487 Hz. The measured frequency spectrum is shown in Figure 1. Figure 1. Noise spectrum at the houses. From the results of the noise survey and the distance to the plant, it was considered that the 487 Hz noise source would be generating a very high noise level at the factory, in excess of 100 dB at one metre, and therefore should be fairly easy to locate. During the on-site noise survey the 487 Hz tone was picked up successfully in the south-western area of a large granulation plant installation. However the relatively low noise level of the tone (87 dB at 487 Hz measured near to a scrubber fan) indicated that the source of the tone was likely to be in the vicinity but higher up on the plant. Following further investigation our interest was drawn not to the main processing plant building, or to the associated maze of external fans, scrubbers, ductwork and piping nearby, but to the top of some very high, but inert looking, silos – an area rarely visited by factory personnel. They were sceptical, 'there's not much up there' we were told. Nevertheless, armed with the necessary permits to climb the gantry up to the top of the silos, we found that each had a small exhaust fan on top and it was these fans that were generating the 487 Hz tone. Each fan unit was found to be emitting the tone at the same intensity. The measured frequency spectrum at a distance of one metre from one of the silo fans, together with the spectrum measured at the nearest residential area, is shown in Figure 2. The correlation of the peak at 487 Hz can clearly be seen. Displaced air from each of the silos is exhausted via a ten blade paddle fan located on top of the silo. Figure 2. Showing the 497 Hz peak from the silo fans. There is a clear line of sight between the top of the silos and the residential area. The primary source of noise emission was found to be a gap between the fan outlet and the cowl of the ducting above. This gap enabled the exhaust noise from the fan to radiate freely to atmosphere. The net result is that there were five high intensity noise sources located in an elevated position with no acoustic screening between the silos and the nearest residential area. Noise Control We were able to give a number of options for reducing the tonal noise ranging from simply installing ductwork between the fan outlet and the cowl above to eradicate the gap, through to replacement of the fans with a different type which do not generate the dramatic blade pass frequency tones which are characteristic of paddle blade fans working at high speeds. In the end, the company opted for fitting silencers to the fan exhausts but, to our consternation, still maintained a small gap between the silencer outlet and the ducting above. Nevertheless, during a follow-up survey it was found that the noise level on top of the silos had been dramatically reduced and, at the houses, the tonal noise had been rendered inaudible. The spectrum measured at the houses after the fitting of the silencers is shown in Figure 3. Figure 3. Spectra at the houses before and after noise control. Therefore the tonal noise problem had been successfully resolved. In order to check that this situation is maintained we are carrying out six-monthly repeats of the environmental and on-site noise measurements. This enables any deterioration in noise levels, and tonal components, to be identified and rectified before becoming a problem. A single instrument, a Svantek 912 Analyser, is now being used for the noise survey work instead of the original monitoring set up. The tone matching technique employed in the case was found to be very successful. It is a useful tool of the acoustic engineer made easier these days by the increasing availability of robust, highly portable, instrumentation which allows real time FFT analysis to be carried out with confidence in the field. Paul Bassett MIOA is a Principal Acoustics Consultant at Hepworth Acoustics Ltd., Warrington.