Long-term environmental noise measurement methods for industrial noise sources

Abstract

Industrial noise sources can impact nearby residential communities, especially in rural areas. In order to assess the situation after commissioning, plant noise must be reliably differentiated from prevailing ambient, transient environmental noises, and weather-related events. Continuously attending such measurements is generally impractical. On the other hand, classifying events by eyeballing a stream of Leq values after the fact isn't sufficiently reliable in most cases either. An effective approach is presented that involves measurement and statistical, observational, and analytical methods to perform much of the work in an automated fashion. Methods include statistically excluding transients from the Leq, filtering to eliminate birds and bugs, spectral characteristics of wind events, and potential spectral identifiers indicative of operation. Classifications are confirmed by listening to periodic and/or event-driven audio segments. This approach works best when both the plant noise and prevailing ambient are quasi-steady; plant-generated transients and weather events may require manual reclassification. Examples from long-term environmental noise studies are provided. The methods are neither dramatic nor novel but their effectiveness in combination may be useful, especially for those new to this type of work.