Low frequency noise and the limits of standard metrics
A-weighting discards about 26 dB at 63 Hz, so plant can pass a rating level assessment and still produce a hum complaint. What to measure instead, and which criteria to report.
A-weighting discards roughly 26 dB at 63 Hz and about 39 dB at 31.5 Hz, so a plant item whose energy sits in the bottom two octave bands can pass a rating level assessment comfortably and still generate a hum complaint within a fortnight of commissioning. That gap between the number and the experience is the whole subject. It is not a flaw in BS 4142, which says plainly that its method has limits where low frequency content dominates. It is a flaw in applying the method without reading that caveat.
The practical question for a reviewer is when to go beyond the rating level, and what to use instead.
Why the weighting understates it
The A-weighting curve approximates the equal loudness contour at around 40 phons, which is a reasonable model for speech-level broadband sound and a poor one for a 50 Hz tone at 45 dB. The correction values are large at the bottom of the spectrum and they are applied before anything else happens in the assessment.
Two consequences follow. A source with a steeply falling spectrum, which includes most large fans and most reciprocating plant, loses most of its acoustic energy on paper. And the character corrections in BS 4142 for tonality and impulsivity are applied to the rating level, which is already A-weighted, so a +3 dB tonal correction on a hum that has been attenuated by 26 dB does not restore the balance.
Complainants describe a low frequency problem differently from a broadband one. Pressure rather than sound. Worse indoors than outdoors. Worse at night, worse in some rooms than others, sometimes felt in the chest. When those phrases appear in a complaint history, the A-weighted figure is already the wrong starting point.
Sources that produce it
The recurring ones in planning casework:
- large axial and centrifugal fans running at low speed, where the blade passing frequency falls below 100 Hz
- air source heat pumps and chillers, particularly in arrays
- transformers and substations, which radiate at twice mains frequency and its harmonics
- combined heat and power units and standby generators
- biomass boilers and their induced draught fans
- amplified music, where the content below 125 Hz carries furthest and is least attenuated by building fabric
- borehole and booster pumps, lifts, and gym drop zones through the structure
The pattern is clear enough. Large, slow, heavy and periodic. If the plant schedule has any of those words against it, scope the assessment accordingly rather than defaulting to a broadband survey.
Screening before you commit
The quickest screening indicator is the difference between the C-weighted and A-weighted level at the receptor. C-weighting is nearly flat down to 31.5 Hz, so a large difference tells you where the energy sits. A difference of more than about 15 dB is commonly treated as a prompt to look at the spectrum properly, and more than 20 dB as a strong one.
That is a screening tool, not a criterion. Use it to decide whether to deploy third-octave measurement, and then stop using it.
| Indicator | What it suggests |
|---|---|
| LCeq minus LAeq below about 15 dB | Broadband assessment is probably adequate |
| LCeq minus LAeq above about 15 dB | Examine the one-third octave spectrum |
| Audible tone identified by complainants but low rating level | Low frequency or infrasonic content likely |
| Complaint worse indoors than outdoors | Room modes or structure-borne path |
Criteria beyond the rating level
Where the spectrum shows the energy below 160 Hz, the assessment needs a band-by-band criterion rather than a single figure. The Defra procedure for assessing low frequency noise complaints, developed for environmental health use, sets third-octave band criteria across 10 to 160 Hz and remains the reference most often cited in England. Several European standards offer comparable band criteria, and some practitioners also present the measured spectrum against the threshold of hearing.
Report more than one. A reviewer should see the BS 4142 rating level, because the authority will expect it, and alongside it a band-by-band comparison with a recognised low frequency criterion. Present them as answering different questions rather than as competing verdicts, and say which one you think governs.
Where entertainment noise is the source, the planning and licensing regimes overlap, and our note on planning control and licensing control of entertainment noise sets out which does what.
Measuring indoors without fooling yourself
Low frequency measurement indoors is where most reports go wrong, because room modes produce spatial variation of 10 dB or more across a single room. A measurement at the centre of a bedroom can sit in a null while the complainant's pillow sits in an antinode.
What a defensible internal survey does:
- Measure at several positions, including room corners, where modal pressure maxima occur.
- Measure in the room the complainant identifies, at the time of day they identify.
- Record one-third octave bands from at least 10 Hz, with a system whose noise floor is demonstrably below the levels of interest.
- Log for long enough to capture the plant cycling, not a five-minute sample.
- Record the source on and source off condition, by arrangement, rather than inferring the residual.
- Calibrate at low frequency and state the microphone and preamplifier noise floor in the report.
Duration and position decide validity in exactly the way they do for external monitoring, and the same discipline applies. Our piece on unattended monitoring duration, position and validity covers the general case.
Structure-borne paths
A hum that is louder indoors with windows closed is usually not coming through the air. Plant bolted to a slab, pipework clipped to a party wall, a duct passing through a structural element: all of them transmit vibration that re-radiates as sound inside the room.
Isolation design is the fix, and it has to be designed rather than specified from a product sheet. Anti-vibration mounts only isolate well above their own natural frequency, so a mount chosen without reference to the forcing frequency of the machine can amplify rather than attenuate. Inertia bases, flexible connections at every pipe and duct penetration, and resilient hangers for suspended runs are the standard measures. Where a reviewer sees "plant to be mounted on anti-vibration mounts" with no natural frequency stated, that is a comment worth making.
Mitigation that works at 63 Hz
Absorption does very little here. The treatments that work are mass, separation and source control.
Attenuators sized for the bottom two octave bands are physically large, often longer than the plant room allows, and manufacturers' insertion loss data frequently stops at 63 Hz, which should make you cautious about any predicted figure below that. Acoustic louvres are close to transparent at low frequency. Barriers work poorly because diffraction performance falls with frequency.
What does work: slower, larger fans with variable speed drives, so the blade passing frequency and the tip speed both drop. Relocating plant away from the façade and from sensitive rooms. Enclosures built with mass and a wide cavity rather than lined panels. Avoiding coincidence between plant forcing frequencies and the structural resonances of the building. And on music venues, a box-in-box construction, because nothing else holds 63 Hz.
Writing the uncertainty honestly
This is the part that separates a usable report from a defensive one.
Prediction at low frequency carries larger error bars than at mid frequency, and the honest report says so, with a figure where one can be justified. State the measurement noise floor. State where the manufacturer's data ends and your extrapolation begins. Where room modes could not be fully characterised, say what that means for the conclusion rather than averaging it away.
A Noise Impact Assessment that sets out a low frequency risk, proposes a design response and recommends a commissioning test is more useful to a case officer than one that produces a single confident number. Conditions requiring post-completion verification exist for exactly this situation. Where an authority wants the work checked by a suitably qualified acoustician before it relies on it, the professional directory is the route to that review.
An illustrative case: a rooftop chiller array above flats, assessed at a rating level 4 dB below background and approved. After commissioning, residents on the top floor reported a hum. Measurement found a 63 Hz band level 12 dB above the Defra criterion in room corners, driven by a structure-borne path through a poorly isolated steel frame. The remedy was mounts selected against the forcing frequency and an inertia base, and it cost more than the original assessment.
Enter your postcode on PlanScope to see which environmental reports a plant installation is likely to need.
PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval. For specialist technical reports, qualified professional input may be required.
Frequently asked questions
Why does a scheme pass BS 4142 and still generate hum complaints?
Because the rating level is A-weighted, and A-weighting removes roughly 26 dB at 63 Hz and about 39 dB at 31.5 Hz. A source whose energy sits in the bottom octave bands therefore scores well on paper. The standard itself notes limits where low frequency content dominates, so the method should be supplemented rather than relied on alone.
What is the C minus A screening check?
C-weighting is close to flat down to 31.5 Hz, so subtracting the A-weighted level from the C-weighted level at a receptor shows where the energy sits. A difference above about 15 dB is commonly treated as a prompt to measure one-third octave bands, and above 20 dB as a strong one. It screens, it does not assess.
Which criteria apply to low frequency noise?
There is no single statutory criterion. The Defra procedure for assessing low frequency noise complaints gives third-octave band criteria from 10 to 160 Hz and is the reference most often used in England. Report it alongside the BS 4142 rating level, present the measured spectrum, and say which comparison you consider governs.
Where should I place the microphone indoors?
At several positions including room corners, because standing waves produce spatial variation of 10 dB or more. Measure in the room the complainant identifies, at the time they identify, in one-third octave bands from at least 10 Hz, with a measurement chain whose noise floor is demonstrably below the levels of interest.
Do acoustic louvres help with low frequency plant noise?
Very little. Louvre and barrier performance falls with frequency, and absorptive lining does almost nothing at 63 Hz. The measures that work are source control through slower larger fans and variable speed drives, mass and separation in the enclosure, relocation away from sensitive rooms, and proper vibration isolation of the plant.
Sources and further reading
This article is general guidance about the planning system in England and, where stated, the rest of the UK. It is not legal or professional advice. Check your local planning authority’s validation requirements and take professional advice for your specific site.