A paper accepted for publication in *Noise & Health* finds that noise from hyperscale data centres – often a low-frequency hum between 70 and 140 Hz – is still not measured accurately enough to assess its effects on health.
Davide Bray and Francesca Domini have gathered direct measurements, regulatory and judicial documents, journalistic surveys and engineering literature on noise from hyperscale data centres built for AI in the United States. Available surveys, taken from riparians, on or near property limits, range from 39 to over 70 dB(A). However, due to the lack of common methods, durations and instruments, these figures cannot be aggregated or compared to regulatory nocturnal thresholds.
A serious sound that conventional measures underestimate
In almost all cases documented with spectral detail, complaints were about a continuous and tonal sound, most often in low frequencies between 70 and 140 Hz, which the usual A weight tends to decrease. The technical sources suggest that the cooling architecture would weigh heavily in the noise emitted, with an estimated difference of 15 to 30 dB between equipment, a hypothesis that the authors still consider to be confirmed in the field.
The article, accepted in the review Noise & Health, proposes a standard measurement protocol, resolved in frequency and applied to named sites: an indispensable basis, in their view, for serious assessment of health effects and environmental justice issues.
Read the full article on the original source
Go to the source →