ACMV Noise Survey: The Measurement a TOP Application Needs

The survey belongs to the Temporary Occupation Permit, and NEA has put that in writing. A 2013 circular from its Central Building Plan Department states that qualified persons must submit a noise assessment report during the TOP application. The plant it covers is air-conditioning and mechanical ventilation, shortened here to ACMV.

That sentence fixes three things: who owes the report, which is the qualified person on the project; the building type, which is non-industrial; and the moment, which is the permit that lets people move in.

Building type decides whether any of this reaches you. The guideline scopes itself to commercial use, drawing its line by what the premises are for: a mall, hotel or office block sits inside; a factory or home sits outside, and each answers to a different instrument. Settle where a mixed development falls before pricing anything.

Landed residential work can drop out. NEA's guidance states that noise assessment is not required for residential landed properties submitted through the lodgement scheme. The owner still carries whatever a badly placed condenser causes later; the submission simply forms no part of the permit.

The common misreading runs the other way. Owners expect a noise survey to arrive with a complaint, triggered by an angry neighbour. Here the trigger is a building reaching completion, and nobody has to be upset for the obligation to exist.

Timing is what catches projects out. The duty attaches to a permit application, and permit applications run to a professional programme. Occupation does not trigger it, a tenant moving in does not trigger it, and a building that never draws a complaint still owes the report.

The report reaches NEA in two stages, and the first happens while the plant is still a drawing. The 2013 circular sets both out in order.

Stage one follows building plan clearance. At that point the qualified person tells NEA where the plant and the monitoring point will go, and NEA assesses those positions before anything is built.

The monitoring location is the line worth pausing on. Where the reading will eventually be taken is agreed on paper, years before a meter appears on site. It follows the boundary and the equipment position, so the design chooses it, not a convenient spot on the day.

Ambient is doing quiet work at stage one. What gets located early is where the surrounding noise will be characterised, and that becomes the reference the plant is later judged beside. Setting it before construction means the reference describes the site as it was.

Stage one is also where the cheap decisions live. Moving a condenser bank on a drawing costs a revision; moving it after commissioning costs pipework, cranage and a delay against a date every other trade is working towards. The later measurement only reports what these earlier choices produced.

Measurement is stage two. It brings follow-up work with it. Where readings fail the current guidelines, the qualified person proposes fixes and then checks how well they work. Fresh readings follow, and so does a resubmission to NEA.

A lead time sits on top of that. The circular requires the information to reach NEA at least 14 days before the TOP application. Treat that figure with care: the circular dates from 2013 and the department named in it has since been renamed, so confirm the current rule. A survey first raised in the closing week of a project is already late.

Four positions are fixed at stage one, and every one of them shapes what the later reading can show.

• •Where the mechanical equipment will stand
• •Where the mechanical ventilation system will run
• •Where the large or centralised air-conditioning system will sit
• •Where the ambient noise will be monitored

Anyone can hold a sound level meter. Only some reports clear quickly, and NEA has built a scheme around that difference.

The Registered Noise Consultants Scheme started on 1 April 2019 to speed up processing of reports prepared under the ACMV technical guideline. Where a registered noise consultant prepares one, NEA states that clearance would be granted within 3 working days.

NEA's reasoning sets out what competence means here. It notes that these consultants can recommend fixes. Their role therefore runs past taking readings.

Entry to the scheme widened from 1 March 2026. A September 2025 circular reset the qualification routes, opening the door to engineering and science diploma holders previously shut out. It also moved the track record test to at least 15 ACMV noise assessments for non-industrial buildings over the past three years.

Two parties sit behind one report, and confusing them wastes time. The qualified person submits it, because the submission belongs to the building plan chain. The consultant taking the readings is engaged commercially, normally by the developer or owner funding the project. NEA's 2025 review puts the split plainly: keeping an active pool of registered consultants available to both.

Two smaller points matter to whoever appoints the consultant. Registered consultants must submit at least one such assessment each year to stay listed, so a name on the register is evidence of current practice. They are also not required to pre-consult NEA about measurement points, removing a step people often assume is mandatory.

Comparing candidates is easier with that list to hand: NEA names what a registered consultant should bring.

• •Technical proficiency in acoustics
• •Thorough knowledge of the technical guidelines
• •Expertise in the appropriate use of sound level measuring equipment

The deliverable is a measured report, judged against boundary noise limits. Those limits shift with the type of premises disturbed and the hour, so no single figure covers every project. Which one applies has its own page, and current numbers should come from NEA for the premises affected.

Three procedural facts shape what the survey costs to get right, all in the technical guideline. The first is that the limit is an average: NEA expresses it as the equivalent continuous noise level over 15 minutes, so a single short bang does not decide the outcome.

The instrument is specified, which rules out the cheap version of this exercise. NEA names IEC 61672 Type 1 and IEC 804 Type 1, or a comparable standard, and expects the meter to be calibrated and integrating. A phone app will produce a number, not a submission.

Reading height follows the equipment, and roof plant is treated differently from plant on the ground. Those mechanics belong to the boundary noise limits page.

Background readings form part of the job, and skipping them leaves the report incomplete. The guideline works a correction from the gap between the permitted level and the background, so both must be captured on the same visit. That also explains why one machine clears a boundary facing a main road and struggles on a quiet back line.

One thing the guideline does not settle is the operating condition. It fixes the measuring point, instrument class and correction factors, but says nothing about what load the plant runs at while the reading is taken. Agree that with the consultant in advance: a reading on a mild morning with half the plant idle describes little about a hot afternoon.

A failed reading does not send anyone shopping for a quieter machine. The remedy in the circular is mitigation, followed by evidence that it worked. That framing matters: it points the spend at the path the sound takes, not at the equipment schedule.

What gets changed depends on how the noise reaches the line. Sound carried through the air suits screening, enclosure or attenuators. Structure-borne sound shows up as a hum indoors, often somewhere the plant cannot be seen, and needs isolation and better mounting. A machine boxed into a recess that throws sound back out may only need repositioning.

Downsizing the plant is the instinct to resist hardest. The equipment was selected against a cooling load, and that load did not change because a reading came back high. A smaller machine trades a noise problem for a comfort problem that surfaces later, reported by a tenant already moved in.

The report keeps working long after the permit is granted. A building holding a measured report holds a baseline: when a complaint lands two years on, the argument stops being about whose hearing is right and becomes a comparison against a recorded figure, taken at a known point.

For a facilities manager the evidence is worth more than the compliance. Noise disputes are hard to settle because both sides are usually honest and neither holds a number: one party hears an intrusion, the other a machine doing the job it was installed for. A dated measurement at a defined point turns that into something checkable before positions harden.

Protection runs in both directions. Where a later reading puts the plant exactly where it always sat, attention moves to what else changed nearby: traffic, a new access road, a neighbouring block finishing its fit-out. Without a baseline, the newest machine tends to get blamed.

That baseline also dates the drift. Plant grows louder with age as coils foul, bearings develop play, brackets loosen and rubber mounts harden. A fresh reading against the original tells an owner whether the neighbourhood changed or the machine did: servicing answers the second case, acoustic work the first.