Low frequency noise complaints and why they are hard

A low frequency noise assessment that reports a single A-weighted level has usually already missed the problem. A-weighting was derived from equal loudness contours at low sound levels and it deliberately discards energy at the bottom of the spectrum: about 16 dB at 125 Hz, about 26 dB at 63 Hz, and close to 40 dB at 31.5 Hz. A plant item producing a strong 50 Hz tone can therefore sit comfortably inside a dBA limit while making a bedroom unusable.

This is the category of complaint that consumes environmental health officers' time and rarely resolves cleanly.

Why A weighting understates low frequency content

The weighting curve is a filter applied to the measured spectrum before the overall level is calculated. It approximates human hearing sensitivity, which really is poorer at low frequencies, and that approximation holds reasonably well for broadband sound at moderate levels.

It breaks down in two situations. Where the sound is tonal and low, because loudness contours crowd together at low frequencies, so a small increase in level produces a large increase in perceived loudness. And where the sound is at or just above the threshold of audibility, which is where most complaint cases sit.

BS 4142:2014+A1:2019 is the usual method for industrial and commercial sound, and it acknowledges its own limits where low frequency content dominates. A reviewer seeing a BS 4142 assessment applied to a complaint about a hum, with no spectral data behind it, should treat that as the finding.

Typical sources in and around buildings

The list is short and repetitive.

Structure-borne transmission is what makes several of these hard. A compressor bolted to a slab excites the building, the building radiates from wall and floor surfaces, and the sound arrives in a room without ever passing through a window. Sealing the window changes nothing, which is exactly the observation a complainant reports and an assessor dismisses.

Third octave measurement and the criteria used

The workable approach measures third-octave band levels, typically from 10 Hz to 160 Hz, over a period long enough to catch the source cycling.

The reference used in England is the Defra-commissioned criterion curve for low frequency noise disturbance, which gives band-by-band thresholds and a method for handling fluctuation. Assessment compares the measured band levels with that curve, inside the affected room, with the complainant absent and the room otherwise quiet.

Points that decide whether the measurement is worth anything:

1. Measure inside the room, at the position the complainant reports, and at several positions in that room.
2. Measure for long enough to cover plant cycling, which for refrigeration can be a full night.
3. Record with the source on and off, if the operator will co-operate.
4. Use instrumentation with a proven low frequency response, confirmed by calibration, rather than a general purpose sound level meter with a rolled-off microphone.
5. Report unweighted band levels, and give the dBA figure separately if it is wanted.
6. Note the fluctuation, because a steady tone and a cycling tone at the same level produce different complaints.

Windscreen and handling noise contaminate low frequency measurements badly. So does the assessor breathing near the microphone.

Room modes and why the level varies within a room

A bedroom is a resonant box. Below a few hundred hertz, sound in a small room forms standing waves at frequencies determined by the dimensions, and the level at any point depends on where that point sits in the pattern.

The practical consequence is a variation of 10 dB or more between the middle of a room and a corner, at the same frequency, at the same moment. A complainant who says the hum is worse in bed, in one corner, with their head against the wall, is describing physics rather than imagining things.

For a reviewer this has two implications. A single measurement position is not a measurement. And mitigation targets that treat the room as a single number will be judged by the occupant at the worst position in it.

Room dimensions also explain why the same source affects one flat and not the one above.

Distinguishing an audible source from tinnitus

This has to be handled carefully and without condescension.

Useful indicators that an external source exists: the sound is present at some times and not others in a pattern the occupant can describe, it changes when the suspected source is switched off, it is audible to more than one person, it disappears when the occupant is away from the building, and it can be masked by a fan or by traffic.

Indicators pointing the other way: the sound is present continuously including away from home, it does not change position within the room, and it has no measurable correlate.

Where measurement finds nothing at the reported time, say so plainly and record the conditions, rather than concluding that nothing exists. Sources cycle. A single unproductive night is evidence about that night.

Mitigation that works at low frequencies

Absorption barely helps. A 63 Hz wavelength is around 5.4 metres, and porous absorbers need depth on the order of a quarter wavelength to do anything, which is not available in a plant enclosure.

What does work, in rough order of effectiveness:

Deal with it at source. Change the plant, change its operating speed, or change the blade or vane count so the tone moves to a frequency the criterion treats differently.

Isolate the structure-borne path. Anti-vibration mounts selected for the actual disturbing frequency, inertia bases with adequate mass, and flexible connections on every pipe and duct crossing the isolation break. A single rigid conduit defeats the whole installation.

Add mass to the airborne path. Mass is the only thing that reliably attenuates low frequencies, which means heavy enclosures rather than acoustic foam.

Distance and orientation. Moving a condenser to the far side of a roof and turning its discharge away is often more effective than any treatment applied where it stands.

Reactive silencers tuned to the offending band, which are large and rarely retrofit into an existing duct.

Writing conditions that capture the problem

A condition expressed only in dBA will not control this. A condition that does the job specifies:

• Limits as unweighted third-octave band levels at defined internal or external positions, referenced to the recognised low frequency criterion.
• A requirement for post-installation verification by a suitably qualified acoustician before first use.
• A requirement that verification is carried out with the plant at its noisiest operating condition, defined in the condition.
• A maintenance obligation, because worn bearings and unbalanced fans generate tones that new plant does not.

Where a Noise Impact Assessment supports an application involving refrigeration, heat pumps or a kitchen extract, the spectral data should be in the report at the outset. The same applies to the plant described in a Ventilation/Extraction Statement, where fan selection is often made after permission and undoes the assessment.

For specialist technical reports, qualified professional input may be required. Low frequency assessment is specialist work needing instrumentation and experience, and a draft report is a starting point for that person rather than a substitute. Reviewers listed in our professional directory can check and sign off drafts of this kind.

PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval. Enter your postcode to see which reports a plant installation is likely to need.