In short. Generator noise on a site is judged in two places. The first is at 1 m from the set, where the factory enclosure rule applies. The second is at the boundary or the neighbour’s window, where the ambient limits apply. Those limits fall from 55 dB(A) by day to 45 dB(A) at night in residential areas. This page gives you a reliable measurement method, a worked background-noise correction, the zone limits, and the order in which to look for the cause when a neighbour complains: exhaust, cooling air path, door seals, then vibration.
Who this page is for: facility managers, apartment association committees, hospital and school administrators, and site engineers who need defensible readings, either to answer a complaint or to prove a site is within limits.
Key facts
- Ambient limits under the Noise Pollution (Regulation and Control) Rules, 2000, as Leq in dB(A), day then night: industrial 75 / 70, commercial 65 / 55, residential 55 / 45, silence zone 50 / 40.
- Day is 6:00 am to 10:00 pm; night is 10:00 pm to 6:00 am.
- A silence zone is an area of not less than 100 m around hospitals, educational institutions, courts and religious places, or any other area the competent authority declares as such. Mixed areas are declared as one of the four categories, and that category’s limits apply.
- The factory limit for the set itself (75 dB(A) at 1 m for sets up to 1,000 kVA), the insertion-loss rule for larger sets and the distance-correction rules are explained on our acoustic enclosure page. This page is about what happens on your site.
- Use a Class 2 or better sound level meter on A-weighting, calibrated before and after with an acoustic calibrator.
- If the set adds less than 3 dB over background, its own level cannot be separated out. The reading proves only that the set is not the dominant source.
Boundary readings versus 1 m readings
A 1 m reading tells you about the machine. Is the enclosure still performing as built, which face leaks most, and has anything changed since last year? It does not tell you whether you are within the law at your boundary.
A boundary or window reading tells you about the place. It is what a neighbour experiences, measured as Leq over a defined period and compared with the zone limit for that time of day. A set can be perfectly healthy at 1 m and still breach the residential night limit at a window 10 m away. The opposite is also true: a tired enclosure may still be acceptable at a distant boundary.
For any dispute you need both sets of readings, taken in the same session, under the same load, with the set on and then off.
Step-by-step on-site measurement
1. Prepare the instrument
- Meter: Class 2 minimum to the current IEC sound level meter standard; Class 1 is preferred for evidence. The meter should integrate Leq.
- Weighting: A-weighting for all compliance readings. C-weighting or octave bands help with diagnosis, because low-frequency hum from exhausts and radiators shows strongly on C.
- Time weighting: Slow (1 s) for steady generator noise and spot readings. Fast (125 ms) only to catch short events such as start-up, breaker operation or rattles. Leq over the period is the figure you compare with the limits.
- Calibration: check with the calibrator before and after the session. If the two checks differ by more than 0.5 dB, repeat the measurements.
- Windshield: always fit it outdoors. Abandon the session if wind is strong enough to move leaves continuously.
2. Set the operating condition
- Run the set at the load it carries during the complaint period. For a compliance check, run it as near rated load as you can, using a load bank if the building load is small.
- Run for at least 15 minutes first, so the radiator fan, which is often thermostatic or viscous-coupled, reaches its working speed.
- Doors closed and latched. Note the kW, voltage and frequency.
3. Choose the positions
- 1 m positions: 1 m from the centre of each face of the enclosure, at about 1.2 m height. That gives four positions, and eight if you add the corners. Stay out of the radiator discharge stream and away from the exhaust outlet.
- Boundary positions: on the boundary line nearest to each affected property, 1.2–1.5 m above ground and at least 3.5 m from walls where possible.
- Receiver position: outside the complainant’s nearest open window, about 1 m from the façade. Record that it is a façade reading, because a façade adds reflection.
4. Measure in pairs
At each boundary and receiver position, record Leq over at least 5 minutes with the set running, then Leq over at least 5 minutes with the set off, taken as close together in time as possible. Traffic, other equipment and birds all change background levels within minutes, so paired readings are the only fair comparison.
5. Correct for background
| Running minus background (dB) | Subtract from running reading (dB) | Confidence |
|---|---|---|
| 10 or more | 0 | High — background negligible |
| 6 to 9 | 1 | Good |
| 4 to 5 | 2 | Fair — repeat at a quieter time if possible |
| 3 | 3 | Low — report as indicative only |
| Less than 3 | Not possible | The set is not the dominant source at that position |
The table rounds the exact correction: set level = 10 × log10(10^(running level ÷ 10) − 10^(background level ÷ 10)), with all levels in dB(A).
6. Record everything
Write down the date, start and end times, meter and calibrator serials, calibration readings, load, weather, positions with distances, and every reading. Add a sketch and photographs of each position. A reading without its conditions will not stand up in a dispute.
Worked example (illustrative): night-time readings at an apartment boundary
The situation. A 160 kVA standby set serves an apartment block in a residential area. The neighbouring house has complained about the weekly test run, held at 10:30 pm. Readings were taken at 10:30 pm with the set on building load of 70 kW.
- Calibration: 94.0 dB before and 94.2 dB after. The drift of 0.2 dB is within 0.5 dB, so the session is valid.
- Boundary position B1 (nearest the house): set running Leq(5 min) = 53.8 dB(A); set off Leq(5 min) = 47.1 dB(A). Difference 6.7 dB, so the table correction is 1 dB.
- Exact correction: 10^5.38 = 239,883 and 10^4.71 = 51,286. The difference is 188,597, and 10 × log10(188,597) = 52.8 dB(A). This agrees with the table.
- Judgement: the generator’s own contribution at the boundary, 52.8 dB(A), is 7.8 dB above the residential night limit of 45 dB(A). The background of 47.1 dB(A) is already above the limit on its own, which is common in a city. The set is still clearly the dominant source while it runs.
- Same position, daytime test at 11:00 am: set running 58.0 dB(A), set off 55.5 dB(A). The difference is 2.5 dB, so the set cannot be separated from daytime traffic. During the day the generator is not the dominant source at this boundary.
- Action: move the weekly test run to mid-morning, which removes the night exposure immediately. Then inspect the exhaust silencer and door seals, following the diagnosis order below. After repairs, B1 at night read running 50.2 and off 47.0. The corrected contribution was 47.4 dB(A), down 5.4 dB but still 2.4 dB above the night limit. Because the set now runs at night only during real power cuts, routine exposure has stopped. A residential-grade silencer upgrade is planned to close the remaining gap.
Lesson: the zone limit and the time of day decided this case more than the set did. Paired readings with a background correction turned an argument into a plan.
When a neighbour complains: the diagnosis order
Work from the largest and most common contributor to the smallest. Most complaints are solved in the first two steps.
1. Exhaust
- Listen and measure near the outlet and along the pipe. A sharp, low-frequency “bark” or a rhythmic pulse points to the exhaust.
- Check silencer condition: rusted-through shells, broken baffles and blown joint gaskets. Soot streaks at joints are a giveaway.
- Check the silencer grade. A basic industrial silencer next to housing is often the whole problem; residential and critical grades exist for a reason.
- Check the outlet direction. An outlet pointing at a building, or into a narrow gap, projects the noise straight at the listener.
- Measure back-pressure after any silencer change, to protect the engine.
2. Cooling air path
- The radiator fan and the hot-air discharge are the second biggest source, and often the loudest by day.
- Check where the discharge points. Turning it away from the complainant, or fitting a discharge hood or attenuator, can remove several decibels on that side.
- Look for recirculation. If hot air loops back to the intake, the set runs hot, the fan runs harder and louder, and staff prop the doors open. That last step undoes the whole enclosure.
- Check that intake louvres are clear. Blocked louvres raise fan speed and turbulence noise.
3. Door seals and enclosure integrity
- Walk around at 1 m and compare faces. A face more than 3 dB louder than the others points to a leak.
- Check door seals for perishing, latches for proper compression, and cable entry gaps.
- Look for missing or oil-soaked acoustic foam, which loses most of its absorption once saturated.
- Check that panels are bolted, not left loose after service work.
4. Vibration paths
- If the complaint is a hum inside a building, rather than at the window, suspect structure-borne noise.
- Inspect anti-vibration mounts for cracking, compression or bolts short-circuiting the mount. Check that the exhaust and cables have flexible connections.
- Terrace and podium installations are the usual offenders. The slab carries vibration into the rooms below, so mount selection and a slab check matter.
- For vibration diagnosis on the set itself, see our vibration and abnormal noise guide.
Operating practices that prevent complaints
| Practice | Why it works | Effect at a residential boundary |
|---|---|---|
| Weekly test runs between 10 am and 5 pm | Uses the 55 dB(A) day limit instead of the 45 dB(A) night limit; daytime background masks the set | Often removes the complaint entirely |
| Tell neighbours the test schedule | Predictable noise is tolerated far better than surprise noise | Fewer complaints at the same level |
| Keep enclosure doors closed; fix overheating instead | An open door can add 10 dB or more | Large reduction |
| Inspect seals and silencer at each service | Catches deterioration before it becomes a dispute | Keeps the site within its original margin |
| Re-measure annually at the same positions | Shows the trend and builds evidence | Defends you in a dispute |
Generator noise measurement record
| Site | Set / kVA | Date | Measured by | Zone category (A/B/C/D) | Applicable limit day / night dB(A) |
|---|---|---|---|---|---|
| Meter make / class / serial | Calibrator serial | Calibration before (dB) | Calibration after (dB) | Weighting / time weighting | Weather and wind |
|---|---|---|---|---|---|
| Load during test (kW) | Voltage / frequency | Run time before readings (min) | Doors closed? (Y/N) | Notes |
|---|---|---|---|---|
| Position | Description and distance from set | Time | Set running Leq dB(A) | Set off Leq dB(A) | Difference (dB) | Correction (dB) | Set contribution dB(A) | Within limit? (Y/N) |
|---|---|---|---|---|---|---|---|---|
| 1 m – front | — | |||||||
| 1 m – rear | — | |||||||
| 1 m – left | — | |||||||
| 1 m – right | — | |||||||
| Boundary B1 | ||||||||
| Boundary B2 | ||||||||
| Receiver R1 (façade) |
| Diagnosis check | Finding | Action |
|---|---|---|
| Exhaust: silencer, joints, outlet direction | ||
| Cooling air path: discharge direction, recirculation, louvres | ||
| Door seals, latches, foam, panels | ||
| Vibration: mounts, flexible connections, slab |
Exceptions and cautions
- Smartphone apps are not calibrated and can be several decibels out. Use them only to spot trends, never as evidence.
- A façade reading includes reflection from the wall behind the microphone. State it as a façade reading; do not compare it unadjusted with a free-field figure.
- Readings at night are affected by temperature inversions, which carry sound further over open ground. Measure at the actual complaint time, not only at a convenient time.
- The zone category is declared by the competent authority. If you are unsure whether an area counts as residential, commercial or mixed, use the stricter limit until you confirm.
- Pollution-board consent conditions or a court order can set stricter limits than the Rules. Follow the stricter figure.
Common mistakes
- Taking one reading with the set running and none with it off, so the background is unknown.
- Measuring at idle, when the fan and engine are at their quietest, and concluding the set is fine.
- Comparing a 1 m reading with the boundary limit, or a boundary reading with the 1 m rule.
- Skipping calibration, or leaving out the instrument details, so the readings cannot be defended.
- Running the weekly test at night, when the residential limit is 10 dB stricter.
- Treating every complaint as an enclosure problem when the exhaust or the discharge direction is the real cause.
What to send us
Send the set’s kVA, a sketch showing the set, the boundary and the complainant’s building with distances, the time the complaint relates to, and any readings you already have. Our service engineers can take paired, calibrated readings at 1 m and at the boundary, work through the diagnosis order, and recommend the repair, silencer, discharge or schedule change that solves it. Use the service enquiry, or WhatsApp 98852 01608. We work across Telangana, Andhra Pradesh and every district within about 300 km of Hyderabad, with project support across India.
Questions engineers ask
What are the day and night noise limits for residential areas?
55 dB(A) Leq by day (6 am to 10 pm) and 45 dB(A) Leq at night (10 pm to 6 am), under the Noise Pollution (Regulation and Control) Rules, 2000.
What is a silence zone and how does it affect my generator?
An area of not less than 100 m around hospitals, educational institutions, courts and religious places, or any other area the competent authority declares, is a silence zone, with limits of 50 dB(A) by day and 40 dB(A) at night. A generator near such premises needs careful siting and a high-grade silencer.
Should I use slow or fast response on the meter?
Use slow response for steady generator noise and spot readings, with Leq over the measurement period for comparison with limits. Use fast response only to capture short events such as start-up or rattles.
How do I subtract background noise?
Take Leq with the set running and with it off at the same position. If the difference is 10 dB or more, no correction is needed. At 6 to 9 dB subtract 1, at 4 to 5 subtract 2, and at 3 subtract 3. Below 3 dB the set cannot be separated from background.
Where should I measure: at 1 m or at the boundary?
Both. The 1 m readings show whether the set and its enclosure are performing. The boundary or window readings show what neighbours experience, and they are what you compare with the zone limits.
What is the usual cause of a neighbour’s complaint?
Most often the exhaust (a worn or basic-grade silencer, or an outlet pointing at a building), then the radiator discharge, then leaking door seals, and finally vibration carried through the structure.
When is the best time to run weekly test runs?
Mid-morning to mid-afternoon, when the higher day limit applies and daytime background masks the set. Avoid test runs after 10 pm.
Related pages
- Acoustic enclosures: the 75 dB(A) rule and insertion loss
- Acoustic engineering for generators
- Generator vibration and abnormal noise
- DG set approvals in Telangana and Andhra Pradesh
- The DG set document file
- Generator site planning checklist
Technical content reviewed by D. Srinivas Reddy, Managing Partner and service engineer. Last updated 5 November 2026.