Cumulative assessment where several plant items are proposed
Assess plant item by item and the total is understated. Here is how to combine sound power levels, choose the worst case period and specify attenuation that works.
A cumulative plant noise assessment that treats each item separately will understate the effect, and it is the single most common technical weakness a reviewer finds in plant noise work. Assess an air source heat pump, two condensers and a kitchen extract fan individually against a background sound level and each may pass comfortably. Combine them and the total can be several decibels higher than the loudest one, which is enough to move a scheme from an acceptable rating level to an adverse one. The assessment has to combine them, and it has to say how.
The arithmetic is not difficult. What goes wrong is the assumptions around it: which items run together, for how long, and what shielding each one gets from the others.
Why item by item assessment understates the effect
Sound levels do not add arithmetically. Two identical sources at the same distance give 3 dB more than one. Ten identical sources give 10 dB more. So a plant schedule with several similar items produces a total well above any individual contribution, and a report that tabulates each item against the background without a combined figure has not answered the question.
The reason it happens is usually procedural. The mechanical engineer supplies sound data item by item, the acoustician assesses what they were sent, and nobody holds the whole schedule. On a small commercial fit-out the schedule may arrive in three separate emails.
Reviewers should treat the absence of a combined rating level as a fundamental gap rather than a presentational one. So should the applicant's own team, because the point of the assessment is to find the problem while attenuation is still inexpensive to specify.
Combining sound power levels correctly
Work in sound power levels per octave or third-octave band where the data allows it, then propagate, then combine at the receptor. Combining overall dBA figures at source and propagating once is quicker and loses the frequency information that determines both barrier performance and the presence of acoustic features.
The logarithmic combination is straightforward: convert each level to its energy equivalent, sum, convert back. Doing it band by band matters because low frequency content from a large fan behaves very differently from the mid-frequency content of a small condenser, and a single dBA sum hides that.
Three data quality points decide whether the result means anything.
- Use manufacturer sound power data at the operating duty, not the nominal rated figure at maximum
- Check whether the quoted figure is sound power or sound pressure at a stated distance, because the two are confused constantly
- Where data is missing, use a reasoned assumption with a stated source and flag it, rather than inserting a plausible number
Duty, standby and what actually runs together
The operating assumption is where assessments are most often optimistic. A chiller plant with three units where two run and one is on standby is a two-unit assessment, provided the control strategy genuinely prevents all three running. If it does not, or if the strategy can be overridden on site, assess three.
Reviewers should ask for the control philosophy in writing. Applicants should offer it, because it is the simplest way to justify the lower figure. Where the standby arrangement is relied on, it is reasonable for the council to secure it by condition, and it is better to propose that wording than to have it imposed in a form that constrains maintenance.
Defrost cycles deserve their own mention. A heat pump in defrost can be louder than in normal operation, and on a cold still night that is exactly when the background sound level is lowest. An assessment based on nominal running conditions misses the worst case entirely.
Worst case hours and simultaneous running
The relevant period is the one with the least favourable combination of high plant output and low background sound. That is not the same as the busiest hour.
| Scenario | Plant output | Background | Usually decisive |
|---|---|---|---|
| Summer afternoon | Cooling at peak | Highest | No |
| Summer night | Cooling reduced | Low | Sometimes |
| Winter night | Heating, defrost cycles | Lowest | Often |
| Early morning | Start-up sequence | Low, rising | Sometimes |
Work through the combinations rather than assuming the summer peak governs. On a residential scheme with air source heat pumps the winter night case is frequently the one that sets the attenuation specification, and it is the one omitted from assessments built around a cooling load.
Where the site also hosts deliveries or servicing, those events combine with fixed plant in the same period, and the analysis in night deliveries and servicing noise at commercial sites belongs in the same worst case table rather than in a separate section.
Directivity and screening between items
Plant items are not omnidirectional point sources, and on a crowded roof they shield each other. Ignoring both simplifications is conservative, which sounds safe until the conservatism drives an attenuation specification that the scheme cannot afford or accommodate.
Directivity is defensible where manufacturer data supports it. A fan discharging vertically presents a different level towards a horizontal receptor than towards a point above it. Screening between items is defensible where the geometry is fixed by the design and secured by drawing reference.
What is not defensible is claiming screening from a plant enclosure that a maintenance engineer will leave open, or from a neighbouring structure outside the applicant's control. State which screening elements are within the development and which are not.
A barrier's real performance depends on the path difference and the frequency content rather than on its nominal height, a point worth reading alongside what an acoustic barrier actually achieves before specifying one across a whole plant deck.
Presenting the combined rating level
The presentation should let a reviewer follow the calculation without rebuilding it.
- Tabulate every item with its sound power level, duty, operating hours and data source.
- State the propagation model and the distances, with a plan showing source and receptor positions.
- Show the individual specific sound levels at each receptor.
- Show the logarithmic combination to a combined specific sound level.
- Apply the character correction for tonality, impulsivity, intermittency and other features, with the reasoning stated.
- Compare the resulting rating level with the measured background sound level for the same period, and state the conclusion in the terms of the relevant standard.
Step five is where judgement enters and where reports vary most. A correction applied without reasoning is an invitation to challenge it. A correction omitted where a tonal fan is plainly audible is worse.
The result belongs in the Noise Impact Assessment, and where the plant serves a kitchen the same schedule usually feeds the Ventilation and Extraction Statement, so the two documents should share one plant table rather than maintaining separate ones.
Specifying attenuation across the schedule
Once the combination is done, attenuation should be specified where it buys the most. Treating the loudest item alone often achieves little, because the remaining items still sum close to the original total. Conversely, a modest reduction applied across several similar items can move the combined figure substantially.
Order the options by decibels per pound and by buildability. Relocation is usually the lowest cost and most effective, followed by selection of quieter equipment at procurement, then local enclosures and attenuators, then barriers, then operational restriction. Operational restriction is last because it is the one the operator will resent and the one most likely to be breached.
One honest uncertainty worth stating in any assessment of this kind: combined predictions carry more accumulated uncertainty than single source predictions, and the commissioning measurement is the real test. Say so, and propose the verification condition yourself.
Assessments of this kind are normally expected to be authored by a suitably qualified acoustician, and professional review is strongly advised. This is general guidance, not legal advice. Check the position with your local planning authority or a qualified professional before relying on it. PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval.
Frequently asked questions
Why does assessing plant items individually understate noise?
Sound levels combine logarithmically, not arithmetically. Two identical sources give 3 dB more than one and ten give 10 dB more. A schedule of several similar items therefore produces a total well above any individual contribution, so a table of items assessed separately against the background does not answer the question the standard asks.
Should sound power levels be combined in octave bands?
Where the data allows it, yes. Combining overall dBA figures at source loses the frequency information that determines barrier performance and the presence of acoustic features. Low frequency content from a large fan behaves very differently from mid-frequency content from a small condenser, and a single dBA sum conceals that.
Can standby plant be excluded from the assessment?
Only where the control strategy genuinely prevents all units running together, and that strategy should be provided in writing and secured by condition. If the arrangement can be overridden on site, assess the full complement. Defrost cycles should also be included, because a heat pump in defrost can be louder than in normal running.
Which time period is the worst case for plant noise?
The period with the least favourable combination of plant output and background sound level, which is often a winter night rather than a summer afternoon. Heating plant runs hard, defrost cycles occur, and the background is at its lowest. Assessments built around a cooling peak frequently miss this case.
Where should attenuation be specified?
Where it reduces the combined figure most, which is rarely the loudest item alone. A modest reduction applied across several similar items often achieves more. Order the options by decibels per pound: relocation, quieter equipment at procurement, local enclosures and attenuators, barriers, and operational restriction last.
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.