Noise assessment for oil and gas facilities: the two problems people confuse
Industrial Noise · SoundPLAN Asia · · 8 min read
A refinery, gas processing plant or petrochemical complex carries two noise obligations at once, and they are not the same obligation. One protects the people working inside the fence. The other protects the people living outside it. They use different quantities, different standards, different receiver positions and different mitigation. Treating them as one exercise is the commonest structural error in noise work on these sites.
Problem one: occupational exposure
Inside the plant, the question is how much sound energy a worker accumulates over a shift. The quantity is the daily noise exposure level, normalised to eight hours. The method is ISO 9612, which sets out three measurement strategies - task-based, job-based and full-day - and, importantly, how to establish the uncertainty of the result.
What makes this hard on a process plant is that exposure follows people, not equipment. An operator on a routine round spends minutes beside a compressor and hours in a control room. A maintenance technician spends a shift inside a single noisy enclosure. Two people on the same site can differ by 15 dB or more in daily exposure while standing next to the same machines at different times.
This is where noise mapping earns its place. A contour map of the plant at working height, built from measured or calculated source data, lets exposure be computed for a defined route and duration rather than estimated from a handful of spot readings. It also makes hearing protection zoning defensible: a posted boundary drawn on a map derived from source data can be explained to an inspector, whereas a boundary drawn from memory cannot.
Problem two: community impact
Outside the fence, the question is the level at the nearest noise-sensitive receiver, usually assessed against a project or regulatory limit that differs by day, evening and night. The propagation method is normally ISO 9613-2, and the assessment is done under conditions favourable to propagation - downwind towards the receiver, or under a night-time inversion.
The sources that dominate here are rarely the ones that dominate occupational exposure. High-frequency sources that make a walkway uncomfortable attenuate quickly with distance and through the air. What reaches a community two kilometres away is low frequency: large air coolers, induced and forced draught fans, flare combustion, compressor casings and, in some conditions, tonal content from rotating equipment. A plant can be perfectly compliant at the fence on an A-weighted basis and still generate complaints because of a 63 Hz tone that the A-weighting largely discounts.
The sources worth modelling separately
- Flares. Broadband, elevated, and highly variable with flow. A flare modelled at normal operation says nothing about an upset condition, which is usually when the complaints arrive.
- Air cooled heat exchangers. Large areas of low-frequency fan noise, often the controlling source at distance, and often the cheapest to mitigate at the design stage and the most expensive afterwards.
- Compressors and turbines. Usually enclosed, so the model is really a model of the enclosure, the ventilation openings and the exhaust - not the machine.
- Piping and valves. Distributed line sources that are easy to omit and can dominate near-field occupational exposure along a pipe rack.
- Fired heaters. Burner and draught noise, with the stack acting as an elevated source.
Why the design stage is the only cheap stage
Noise mitigation on an operating plant is retrofitted around live process equipment, under permit, often at a shutdown. The same mitigation specified at layout stage is a line in a purchase order. The difference in cost is not marginal, it is categorical.
The practical consequence is that a plant noise model is most valuable before the plot plan is frozen. At that point the levers are still free: where the air coolers sit relative to the boundary, which direction the compressor house openings face, whether the control room is on the quiet side, whether a bund or building can be positioned to screen the controlling receiver. Once steel is set, the remaining levers are acoustic enclosures, silencers and lagging - all of which work, and all of which cost.
Keeping the two assessments honest
The two problems share one thing: the source data. A single set of octave band sound power levels, properly established, feeds both the internal exposure map and the external propagation model. Where projects go wrong is maintaining two disconnected source lists - one from the vendor data sheets for the environmental study, one from a survey for the occupational study - which then disagree, and nobody can say which is right.
Build one source inventory. Record for each source whether the level is measured or from vendor data, whether it is octave band or a single figure with an assumed spectrum, and what operating condition it represents. Both assessments then rest on the same evidence, and when a regulator or an insurer asks where a number came from, there is an answer.
Industrial noise assessment · Occupational noise mapping · Discuss a plant assessment