Sound Radiation of a Smoke-Removal Fan System

Noise Generation Mechanisms

To remove toxic fumes from people caught in fire, firefighters use more and more powerful fan systems. These machines are designed to provide the maximum flow rate when the fire is the most intense. This ventilation power is directly connected with the problem of sound nuisances that interfere with the firefighter communications even if helmets are worn. The noise also causes premature fatigue for the firefighters. On a more long term, this sound nuisance (mainly the high-pitched tonal noise at the blade passing frequency and its harmonics) can trigger losses of hearing and eventually deafness, or other health hazards for the firefighters (increased heart diseases for instance). The present study is meant to provide details characterizations of the radiated sound of such systems and identify ways of reducing the associated noise.
A representative fire fan system has then been studied in the recently-built anechoic wind tunnel at Université de Sherbrooke at several operating conditions. It consists of a thermal gas engine connected to a fan. The engine is a 6 Hp Honda running from about 1440 RPM at idle to 3660 RPM at full speed. The fan itself is protected by a front and a rear grill, both attached to a volute. The fan tip clearance is more than 1 cm for safety purpose. The size of the fan is 21 inches in diameter. The fan has 7 blades. It can be tilted from 0 to 20 degrees to blow the air slightly upward. Far-field sound pressure spectra and directivity have been obtained. Three different configurations have been studied and compared: the complete fan system, the fan system without the front grill, and the fan without front grill and volute. The microphone directivity array consists of 19 Bruel and Kjaer (B&K) microphones. Excellent repeatability is obtained on the broadband noise (maximum variation of 0.5 dB) and a good one on the tonal noise (maximum variations of 2-3 dB). Scaling laws of the power spectral densities of the far-field acoustic pressure with fan speed have been deduced. Moreover noise source localization has been achieved with two different methods. Conventional beamforming technique is compared with a new de-dopplerization technique for rotating sources. The source localization array consists of a 60 microphone logarithmic spiral array.
Finally having shown the main dipolar noise feature of sound of the fire fan system and the respective contribution of the thermal engine and the fan, a method to separate rotating and stationary noise sources has been successfully applied. The contribution of both noise components are then clearly identified, which allows a separate analysis of both noise sources. Conclusions are finally drawn and suggestions for future noise control are provided.