Finding the Perceived Noise

Sondare recently performed a noise study to determine the source of a problematic noise coming from an aquaculture (fish farm) operation. Two previous studies were performed and both determined that the operation was producing low frequency noises attributed to several pumps running. These studies also provided data showing some noise in the 2000Hz to 3000Hz range that was attributed to birds.

I started my analysis with a walk around the pond and dock area and could detect sounds from the pumps located on the docks. I also used a test meter to analyze the frequency spectrum of the surrounding sounds. I detected some low frequency noise but also detected a significant frequency component around 2500Hz. It matched up to the “bird” noise documented in previous studies.

I did not think this was a coincidence so I investigated further. I recorded several three to five minute audio samples and using a Fast Fourier Transform (fft) calculation in Matlab, I analyzed the frequency spectrum of the audio samples. The results were very interesting. I found six unique frequencies in the 2500 Hz range with different amplitudes. What was interesting was that there were six aerator pumps running. Since the audio sample was only about 50 feet away from the nearest pump the amplitudes were significantly above the ambient noise level. Also since the pumps varied in distance from the audio recorder the amplitudes of each frequency component were all different.

My next step was to see if there was a correlation between the measured frequency components and the electrical and mechanical properties of the aerator pumps. In looking at the spec sheets each aerator pump had 45 blades and ran at 3450 RPMs. The motor shaft frequency was calculated as fm = RPMs/60 cycles/sec = 57.5 Hz. The sound of six motors producing this frequency of sound contributed to the measurement of low frequency noise. But this did not explain the 2500Hz noise.

Due to the spinning fan blades, I calculated the fan blade noise resulting from the fan tip passing the surrounding pump housing. The fan blade frequency (fb) can be calculated as fb= RPMs x no. of blades/60. This calculation produced a fan blade tip noise of 2587Hz. The six frequency components I measured ranged from 2546 Hz to 2562 Hz. The variation from the calculated fan blade noise could be attributed to variations in RPMs of each motor. Although the frequency components did not exactly match the calculated value, I was confident that the fan blade tip noise was the source of the annoying noise being heard.

As a follow up to the analysis I requested that the pumps be turned off. Using the test meter I observed a decrease in amplitude as each pump was individually turned off. When all six pumps were off there was no longer a frequency component at 2500 Hz. This was a final confirmation that the noise was originating from the six aerator pumps.

Now that the source of the noise and the frequency components were verified, a recommendation was made to build a passive absorber enclosure to cover each pump in order to reduce the aerator pump noise. The enclosure should be designed to maximize absorption of sound in the 2000-3000 frequency range. This could be accomplished by using absorption panels made for outdoor use with a high absorption factor in this frequency range or by designing a Helmholz resonator panel enclosure tuned to this frequency range. In either case consideration needs to be given to providing adequate ventilation for the pumps to keep them running cool and not shorten their life span.

This was a great opportunity to use detailed frequency spectrum analysis tools to find the exact cause of a noise problem. Without a detailed analysis, noise problems can be easily attributed to sounds that naturally occur in the outdoor environment such as birds.

If you are in need of solving a noise problem, whether indoor or outdoor, please contact us for an appointment.