DG Set Noise Limit in Your Society: Rules Every Society Should Know

Every conversation about DG noise in an Indian apartment complex arrives at the same sentence, usually from the facility manager or the AMC contractor:

“Our genset is a 75 dB CPCB-compliant model.”

That statement is often true. It is also almost entirely irrelevant to whether your complex is complying with the law. The DG set noise limit that everyone quotes is not the one that governs your building, and understanding why is the most useful thing a Managing Committee can learn about this subject.

Two completely different numbers

75 dB(A) at 1 metre is a product standard. It comes from Schedule I, serial no. 94 of the Environment (Protection) Rules, 1986, notified by G.S.R. 371(E) dated 17 May 2002 and amended through to G.S.R. 215(E) of 15 March 2011. It says:

“The maximum permissible sound pressure level for new diesel generator (DG) sets with rated capacity upto 1000 KVA, manufactured on or after the 1st January, 2005 shall be 75 dB(A) at 1 metre from the enclosure surface.”

That limit binds the manufacturer. It is policed at the factory through Type Approval and Conformity of Production certificates. It is why the machine carries a conformance label reading “75 dB(A) at 1 m” along with its Type Approval number and date of manufacture.

(Incidentally: go and read that label. It is on the set, it is required to be there, and most committees have never looked at it. It tells you the date of manufacture, which tells you which rules apply.)

The limit that applies to the area your residents live in is a different number entirely. The Noise Pollution (Regulation and Control) Rules, 2000, S.O. 123(E) dated 14 February 2000, as amended to 10 August 2017, set ambient standards by zone:

Rule 4(1) is unambiguous: “The noise levels in any area/zone shall not exceed the ambient air quality standards in respect of noise as specified in the Schedule.”

So a DG set can be entirely compliant at 1 metre and still put the area around it outside the ambient standard. A complex cannot defend itself by pointing at the machine’s certificate. The certificate is about the machine. The 45 dB(A) is about the area people sleep in.

One thing to check before relying on it: zone categorisation is done by the State Government, and a mixed area can be declared into any of the four categories. Confirm how your area is actually categorised.

So what actually binds your association?

Here is where most writing on this subject, including, until it was checked, an earlier draft of this article, goes wrong.

The same CPCB entry contains a second set of obligations: an acoustic enclosure or acoustically treated room at the user’s end, designed for a minimum 25 dB(A) insertion loss, plus an exhaust muffler of at least 25 dB(A).

Those are real requirements. But read the heading they sit under:

“2. Noise limit for DG sets not covered by paragraph 1.”

Paragraph 1 is the ≤1000 kVA, post-January-2005 category. So paragraph 2, including the 25 dB(A) insertion-loss and muffler requirements, governs the sets outside that: bigger than 1000 kVA, or manufactured before 2005.

If your complex was built in the last twenty years, your genset is almost certainly a paragraph-1 set, and those two 25 dB(A) numbers are not imposed on you by that entry. Anyone who quotes them at you without checking the heading has not read the page. It is worth knowing that before you take advice from someone who has.

What does reach your installation is paragraph 6. Verbatim:

“6.0 Transportation of Diesel Generator Sets (above 250 KVA) … ii) Compliance with the noise norms shall be monitored after reassembling the DG set at the location of the installation by the concerned State Pollution Control Board, or as the case may be, the Union Territory Pollution Control Committee.”

That is the provision that matters, and almost nobody cites it. For any set above 250 kVA, which covers most apartment complexes, compliance is to be monitored on your site, after installation, by your State Pollution Control Board. Not at the factory. Not on the certificate.

There is also paragraph 2.5, which is guidance rather than a binding limit. Read it with the same care as the rest of paragraph 2: like 2.4, it sits under the heading “Noise limit for DG sets not covered by paragraph 1”, so on its face it is addressed to the bigger and older machines. We quote it because it is revealing about what the regulator expects of an operator generally, not because it imposes a duty on a paragraph-1 set:

“02. The user shall make efforts to bring down the noise levels due to the DG set, outside his premises, within the ambient noise requirements by proper citing [siting] and control measures.”

“04. A proper routine and preventive maintenance procedure for the DG set should be set and followed… which would help prevent noise levels of the DG set from deteriorating with use.”

Read that last clause again. The regulation itself anticipates that a DG set gets louder over time unless somebody maintains it against that. It is written into the rule, as guidance rather than a duty, and in a paragraph aimed at other machines, but written in all the same. Treat it as the regulator’s view of good operating practice, which is how we use it here.

If your AMC scope is engine servicing only, oil, filters, coolant, then nobody is inspecting the acoustic condition of the enclosure. Check your scope.

One myth worth killing: CPCB IV+ did not make gensets quieter

You will be told, when a new set is being sold to your complex, that CPCB IV+ compliance addresses noise.

It does not.

CPCB IV+ was notified as G.S.R. 804(E) dated 3 November 2022 (the Environment (Protection) Third Amendment Rules, 2022), in force from 1 July 2023 and amended by G.S.R. 436(E) of 14 June 2023. It amends Schedule I entries 88, 95, 95A, 95B and 95C. It does not touch serial no. 94, the noise entry, and the word “noise” does not appear in it at all. Its content is emission limits: NOx, hydrocarbons, carbon monoxide, particulate matter, smoke opacity.

The DG set noise limit in India is still the 2002 standard, unchanged since 2011.

One nuance, so you are not caught out: CPCB IV+ certification does bundle a noise certificate, the conformance label carries Type Approval for emissions and noise. But the noise limit it certifies against is unchanged: still 75 dB(A) at 1 m, still serial no. 94, still the 2002 standard.

A CPCB IV+ set is not a quieter set. It is a cleaner set, certified against the same noise number.

The number nobody quotes: what a flat is supposed to sound like inside

The two numbers this article started with, 75 dB(A) at the machine, 45 dB(A) in the area at night, are both outdoor numbers. Neither tells you what is an acceptable level inside somebody’s flat. The national standard does say something about that, and almost nobody quotes it.

NBCS 2026 (SP 7:2026), Part D ‘Building Services’, Section 4 ‘Acoustics, Sound Insulation and Noise Control’, Table 4 ‘Acceptable Indoor Noise Levels for Various Buildings’, clause 4.1, printed page 18: Apartments, hotels and homes, 35 to 40 dBA

SP 7:2026 is a free download from the BIS portal, check the table yourself.

But the same code carries a second table, in a different section of the same Part, that gives a different answer.

NBCS 2026 (SP 7:2026), Part D ‘Building Services’, Section 3 ‘Air Conditioning, Heating and Mechanical Ventilation’, Table 1 ‘Requirements of Acoustic Comfort’, clause 3.4.8.1, printed page 19: Residences, apartments, condominiums, Noise Criterion Class A 30, Class B 35, Class C 40 dBA

and clause 3.4.8.1 itself: “Class A refers to aspirational, Class B refers to acceptable and Class C refers to minimum acceptable.”

So one table gives a home 35 to 40 dBA, and the other grades the same home 30 / 35 / 40 depending on whether you are aiming at aspirational, acceptable or minimum acceptable. NBCS does not reconcile the two, and we are not going to pretend that it does. What a committee can take from it is the band the national standard is working in, 30 to 40 dBA indoors, depending on which table and which grade you read, which is far below either of the outdoor numbers everyone argues about.

Two things the code does not give you, and you should not let anyone tell you it does:

• There is no day/night split. Both tables give a single figure. Unlike the ambient standards outdoors, the indoor figure does not tighten at night.
• There is no separate bedroom figure. The row reads “apartments, hotels and homes”, a dwelling, not a room type. (The WHO bedroom value of 30 dB, quoted later in this article, comes from a different document with a different purpose. Do not merge the two.)

And a warning about a table that will be quoted at you: NBCS Part D Section 4 also carries a Table 5 on sound insulation between individual rooms. Every row in it is flat-to-flat, dwelling to dwelling, room to corridor. It does not govern a wall between a dwelling and a plant room, and it should not be used in an argument about a generator.

BIS does name the diesel generator, under the residential heading

Not prominently, and with no number attached, but it is there:

NBCS 2026 (SP 7:2026), Part D ‘Building Services’, Section 4 ‘Acoustics, Sound Insulation and Noise Control’, clause 5.2.3.2 ‘Suppression of noise at the source itself’, under ‘5 RESIDENTIAL BUILDINGS’, printed page 20: “Special noise control measures may be required for electrical and mechanical services such as diesel generators, outdoor air conditioning units, cooling towers, etc.”

Note the verb. “May be required” is not “shall”. This is the national standard acknowledging that a residential building with a diesel generator may need noise control beyond the ordinary; it is not the national standard ordering anyone to do anything. Quoted honestly it is still useful, it establishes that the people who wrote the code treated a DG set in a residential complex as a special case, which is more than most facility managers will concede.

Where the generator is supposed to go, and which standard says so

There is a claim implicit in most writing on this subject, and an earlier version of this article carried it too: that there is no Indian code for installing a generator, only a rule about how loud it may be. That is wrong, and the correction is useful to a committee.

Siting

NBCS 2026 (SP 7:2026), Part D ‘Building Services’, Section 2, clause 4.3.1 ‘Location’, printed page 40: “The generating set should preferably be housed adjacent to LV switchgear in the substation building to enable transfer of electrical load efficiently and also to avoid transfer of vibration and noise to the main building.”

That is the code naming, in its own words, the exact failure this article describes: vibration and noise transferring into the building people live in. Note that it is written as a preference, “should preferably”, not as an obligation.

And it is the whole of what NBCS gives you on siting. There is no numeric setback distance anywhere in NBCS, no “so many metres from the nearest habitable window”. The code’s Annex D sizes the generator room by kVA; it does not place that room at a distance from anything. If a consultant tells you the code requires the DG yard to be a stated number of metres from the tower, ask them for the clause number.

On the enclosure itself, the code hands the question straight back to CPCB:

NBCS 2026 (SP 7:2026), Part D ‘Building Services’, Section 2, clause 4.3.3(c), printed page 40: “Acoustic enclosure for DG sets/acoustic lining of the DG room and ventilation system for DG room shall be in line with the requirements of CPCB.”

The installation code almost nobody cites

The generator’s installation has its own national code, and it is not the National Building Code.

One thing to be straight about before quoting it. SP 30 is a BIS Special Publication, and by its own words in the clause below it “provides guidelines”, the same standing we give SP 35 elsewhere in this series. It is not a specification that anyone is automatically in breach of. It matters for a different reason: it is the document a competent consultant, contractor or AMC is expected to be working to, and NBCS 2026 points at it by name. Treat it as the benchmark to ask questions against, not as a rule to accuse anybody of breaking.

SP 30:2023, National Electrical Code of India (2nd Revision), Part 2 Section 1, ‘Electrical Installation of Stand-by Generating Stations’, clause 1 (Scope), printed page 388: “This section of the code (Part 2/Sec 1) provides guidelines in respect of electrical design, selection, along with installation, testing, commissioning, maintenance and statutory compliances in respect of standby generating sets, meant to supply power during the period of supply interruption of primary source, or during the emergency, having individual capacity up to and including 2000 kVA.”

“Up to and including 2000 kVA” covers essentially every apartment complex in the country. And inside that section sit precisely the items this article’s remedy ladder is built on: siting factors (clause 6.1), foundation and anti-vibration mounts (clause 6.3.2), and the silencer (clause 6.5.5). On acoustic enclosures:

SP 30:2023, Part 2 Section 1, clause 5.6.1 ‘Noise pollution – Acoustic Enclosures’, printed page 392: “Under CPCB norms it has been made mandatory to maintain noise limit 75 dB* at 1 m from enclosure or surface. For OEM it is mandatory provide Diesel Generating Set with integral acoustic enclosure for capacity up to 800 kW/1000 kVA (see also IS/ISO 8528-10).”

(That is reproduced exactly as printed, the missing “to” and the asterisk are in the original. We do not silently correct the documents we cite, because a quotation that has been tidied up is a quotation that cannot be checked.)

Note what SP 30 does not do: it sets no noise limit of its own. It repeats CPCB’s, and CPCB’s, unlike SP 30’s guidance, does have legal force behind it through the Environment (Protection) Rules. When you quote the 75 dB figure at anybody, quote it as CPCB’s requirement, not as SP 30’s. And note clause 6.1, its list of siting factors includes the distance of the gen-set from the building, and gives no number for it. Two national documents say the distance matters; neither says how much.

The honest summary

There is no single BIS standard for installing a diesel generator, and no single DG set noise limit buried in the building codes either. The requirements are split across four documents, and the one number everybody argues about is in none of them:

We hold NBCS 2026 and SP 30:2023 and have read the clauses quoted above. We do not hold IS/ISO 8528, IS 13301 or IS 2974, so nothing in this article is quoted from them, they appear here because NBCS and SP 30 name them, and because a committee should know which document to ask its consultant about.

The chain that joins the first two rows together is worth knowing, because it is how a building-code requirement becomes an installation requirement: NBCS 2026 Part D Section 2, clause 4.3.3(g) requires compliance with good practice [D-2(1)], and entry (1) in that section’s own list of standards is SP 30:2023.

The practical use of all this. This article has already told you to check your AMC scope. Here is the sharper version of that instruction: ask whether your AMC references SP 30:2023 Part 2 Section 1. Most engine-only maintenance contracts do not, and that is exactly why the acoustic condition of the installation, which is the thing that actually degrades, appears on nobody’s checklist.

Why it got louder, the ten-year sequence

Residents usually report this as a sudden change. It rarely is. Here is the pattern, and the reason a complex often crosses two thresholds at once.

Years 0-2. New set. Enclosure intact, gaskets soft, acoustic foam clean, anti-vibration mounts at design stiffness, silencer new. Complaints are rare and are usually about siting, the flat directly above the yard, rather than about degradation.

Years 2-4. Silent decay. Door gaskets take a compression set. Foam absorbs oil mist and dust and loses absorption. Mount elastomer creeps and stiffens. The change is one to three decibels, below the threshold at which most people notice anything at all.

Years 3-5, the hinge. The radiator core and the ventilation attenuators foul with dust and lint. Airflow drops. On a long summer night run, the set overheats and trips. Under pressure to restore power, somebody opens the enclosure doors or unbolts side panels.

It works. And it becomes permanent, because nobody re-instates them and nobody ever connects the two facts.

This single event is usually the largest step change in noise the complex will ever experience, an enclosure with a panel removed is, on that face, not an enclosure. It is also why the complaint tends to arrive suddenly rather than gradually.

Years 4-7. The AMC moves from the OEM to a cheaper local contractor. Scope narrows to engine servicing, oil, filters, coolant, and drops enclosure integrity entirely, because acoustic condition appears on no checklist and in no log. Silencer internals corrode. The exhaust flexible bellows fatigues and gets replaced with rigid pipe by a fabricator with no reason to know why it was flexible.

Years 5-8. The mounts are now hard or cracked. And the complaint changes character: residents stop saying “it’s noisy” and start saying “I can feel it,” “the windows rattle,” “it’s a hum I can’t get away from anywhere in the flat.”

That shift is diagnostically important. It means the problem has moved from airborne to structure-borne, and the remedy is no longer the enclosure.

Years 6-10. Occupancy hits 100%. Air-conditioning and EV charging load grows. Run hours rise. Sometimes a second set is added, sited wherever there was space. Meanwhile the residents who had accepted the DG yard as a known trade-off have moved on, and the new owners never agreed to it.

So it did not get louder in one step. It degraded quietly for years, took one large step at a thermal event, and then crossed a perception threshold at the same moment the tolerance threshold moved. Both have to be addressed.

The test that decides where your money goes

Before spending anything, establish whether the problem is airborne or structure-borne. Getting this wrong wastes the entire budget, money spent refurbishing an enclosure does nothing for a complaint that is travelling through the building’s structure.

The free version, which any resident can do: in the affected flat, with the DG running, measure or simply listen with the windows fully open, then fully closed.

• Big drop when the windows close → airborne. The path is through the air, and the enclosure, the ventilation apertures and the exhaust are where to look.
• Little or no drop → structure-borne. The vibration is coming through the building. Look at the anti-vibration mounts, at rigid connections bypassing them, and at how the DG plinth is tied into the structure.

The instrumented version: an accelerometer on the flat’s floor slab or a shared column, with the DG running and stopped. Correlating the vibration spectrum with the audible tone at the engine’s firing frequency and its harmonics is the definitive test. A 1500 rpm set turns at 25 Hz; for a four-stroke its firing frequency is 25 × (number of cylinders ÷ 2) Hz, 75 Hz for a six-cylinder, 100 Hz for an eight. Look at both, and at the harmonics.

There is a related test almost nobody runs and it is cheap: put an accelerometer on the DG skid above the mounts and immediately below on the plinth. The ratio tells you the actual in-situ performance of your anti-vibration mounts. If the level below is close to the level above, the mounts are doing nothing, and you can say so with data instead of opinion.

If the complaint is worse on the seventh floor than on the second, follow it up, but do not treat it as proof on its own. Structure-borne noise does not fall off with distance or respect line of sight, so elevation is a genuine clue. But upper floors also have clearer line of sight over any barrier, so airborne noise can rise with height too. The windows test and an accelerometer above and below the mounts are what actually separate the two.

There is no vibration limit to measure against, and we want to be exact about that claim

Once a committee accepts that the complaint is structure-borne, the next question is always: how much vibration is too much? There is no Indian answer to that question, but the sweeping version of that statement is refutable, so here is the precise one, along with the material someone will quote at you.

NBCS 2026 does contain vibration numbers. Three of them. None of the three governs machinery vibration entering a dwelling:

• 0.18 m/s² for residential buildings, NBCS 2026, Part C, Section 5, Annex E, Table 10, printed page 42. This is wind sway of tall reinforced-concrete buildings: the building moving in the wind, not a machine shaking it.
• 5 / 10 / 15 mm/s peak particle velocity, NBCS 2026, Part C, Section 1, Annex L, Table 50, printed page 187. This is blast vibration, and what it protects is the structure against damage, not the resident against disturbance.
• A “0.5 percent g” threshold, NBCS 2026, Part C, Section 6A, Annex C, printed page 184. This is footfall vibration of floors, people walking.

And BIS does publish on the other side of the problem: IS 13301:1992, Vibration Isolation for Machine Foundations, Guidelines, and IS 2974 on machine foundations, and NBCS 2026 Annex K (Part C, Section 1, invoked by clause 8.7 ‘Vibrations’; printed pages 166–167) tells a designer to check human comfort. (We hold neither IS standard and quote nothing from them.)

So the accurate claim is narrower than the one usually made, and it is stronger for being narrow: no Indian standard or rule says how much vibration a resident may be subjected to from machinery, even though standards exist on isolating the machine. You can be told to isolate the foundation. You cannot be told what level inside the bedroom counts as a failure.

The practical consequence for a committee is that on the vibration side you cannot win the argument by citing a limit, because there is no limit to cite. What you can do is the test described just above: measure on the skid above the mounts and on the plinth below, and show that the isolation is not doing its job. That is a comparison against the installation’s own design intent, and it does not need a national number to be persuasive.

Why “we passed the test” and “residents are still complaining” can both be true

The law is written in dB(A). A-weighting deliberately discounts low frequencies to approximate how human hearing responds at moderate levels.

But DG noise is low-frequency dominated, exhaust pulsation at firing frequency, engine block radiation, structure-borne rumble. A-weighting discounts a 31.5 Hz octave band by about 39 dB, a 63 Hz band by about 26 dB, and a 125 Hz band by about 16 dB.

The practical consequence is that a complex can be genuinely, measurably compliant in dB(A) and residents can still be miserable, because the energy that is bothering them sits low, where the measurement counts it least, where barriers are ineffective, and where ordinary glazing gives you almost nothing.

A competent measurement therefore captures both A- and C-weighted levels, and ideally third-octave band data, because that is what tells you whether you are chasing the exhaust, the radiator fan, or a broadband leak. Chasing the wrong one is expensive.

What the measurement has to look like to be worth anything

If your complex is going to spend money on the basis of a noise report, or use one in a complaint, it has to meet the method CPCB actually prescribes. The reference is the CPCB Protocol for Ambient Level Noise Monitoring, July 2015:

• A Class 1 / Type 1 integrating sound level meter. The protocol specifies IEC 61672-1 Class 1 or equivalent. Class 2 is not adequate for a compliance-grade report, and a phone app is worthless, it will be dismissed. There is an Indian designation for the same specification, and quoting it closes off one line of argument: NBCS lists the meter standard as “IS 15575 (Part 1) : 2016/ IEC 61672-1 : 2013 Electroacoustics, Sound level meters, Specifications (first revision)” (NBCS 2026 (SP 7:2026), Part D ‘Building Services’, Section 4 ‘Acoustics, Sound Insulation and Noise Control’, List of Standards, entry (4), printed page 74). We do not hold IS 15575 itself, so nothing is quoted from it. Two independent sources agree on the designation: NBCS lists it as above, and the BIS catalogue record for IS 15575 : Part 1 : 2016 (IEC 61672-1 : 2013), Electroacoustics, Sound Level Meters Part 1 Specifications (First Revision), technical committee LTD 7, shows it active and reaffirmed in 2022. So asking for “a Class 1 meter to IS 15575 (Part 1)” is asking for exactly what the CPCB protocol requires, in Indian terms.
• Field calibration before and after, agreeing within 1.0 dB, with a certificate traceable to national standards.
• Tripod mounted, microphone 1.2–1.5 m above ground. The protocol says hand-held monitoring should be avoided.
• Wind under 5 m/s, windshield fitted, no measurement in fog or rain.
• Leq, L10, L50, L90, Lmax and Lmin at 1-second sampling.
• Night measurement is not optional. The night limit is 10 dB tighter, the background is far lower, and sleep disturbance is the actual mechanism. CPCB’s own DG entry says insertion loss performance may need checking “preferably, in the night time.”
• Background correction, measure with the DG off, then on, at identical positions. This is the step most commonly skipped and it is what separates a report from a number. If the difference between DG-on and background is under 3 dB, the source is not reliably separable and an honest report says so rather than inventing a figure.

And the positions matter. Measure at each face of the enclosure, and separately at the air inlet, the radiator discharge and the exhaust termination, because the objective is to rank the apertures, not to produce one number. Then at the affected flats, at several heights. Measuring only at ground level is a common and serious error: upper floors frequently have direct line of sight over any barrier and can read higher than the ground.

The remedy ladder

Work down it in order. The cheap items are not consolation prizes, they frequently deliver a large share of the achievable improvement, and doing them first is how you avoid spending capital on the wrong thing.

Free / operational

• Log every start, stop and duration. Most complexes cannot state how many night hours their set runs. You cannot manage what you do not measure, and the log is also your evidence.
• Shed non-essential load at night so the set runs fewer hours. Where there are two sets, run the smaller one at night.
• Never schedule the weekly test run at night. Move all DG-dependent maintenance out of 10 p.m.–6 a.m. Free, and it removes the most inflammatory events.
• Close the doors. Obvious, routinely violated, and if they are open because the set overheats, that is a fault to fix, not a workaround to accept.
• Clean the radiator core and the ventilation attenuators. This addresses the root cause of panel removal. Restore the airflow and you remove the reason the enclosure gets opened.
• Service the engine to specification, injectors, valve lash, air filter, governor calibration. Fixes knock and hunting.
• Publish the run-hour log and the findings to residents. A large part of the perceived severity of a DG complaint is loss of trust rather than decibels.
• Ask whether your AMC references SP 30:2023 Part 2 Section 1, the installation code described earlier, which covers siting, mounts and the silencer. If it does not, nobody under that contract is looking at the acoustic condition of the installation, and nobody is being paid to.

Low cost

• Replace all door and hatch gaskets; adjust latches and hinges so doors actually compress the seal.
• Seal every penetration, cable glands, conduit, fuel and drain lines.
• Re-instate missing panels (after fixing the overheating, or they will come off again).
• Replace degraded acoustic foam and corroded perforated inner sheet.
• Fit or replace the exhaust flexible bellows, cheap, and it stops the exhaust short-circuiting the isolators.
• Repair or replace corroded silencer and tailpipe sections.

Medium

• Replace the anti-vibration mounts, correctly specified, chosen for the dynamics, not merely sized to hold the weight. And remove every rigid bypass at the same time: one rigid exhaust pipe, fuel line, conduit or hard-grouted bolt defeats the entire isolation system regardless of how good the mounts are.
• Upgrade the exhaust silencer. Be careful how this is justified. The 25 dB(A) muffler figure people quote sits in paragraph 2 of the CPCB entry, the paragraph that, as set out earlier in this article, governs sets outside the ≤1000 kVA post-2005 category. If your set is a paragraph-1 set, that number is not imposed on you by that entry, and presenting it to your committee as a legal requirement is the same mistake this article warns about. Treat a silencer upgrade as an engineering decision justified by measurement, not as a compliance item, unless you have first established that your set falls under paragraph 2.
• Re-route and raise the exhaust termination, above the affected façades, aimed away from towers.
• Full enclosure refurbishment, including refurbished or correctly sized ventilation attenuators. Once the doors are sealed, the ventilation apertures are usually where the remaining decibels are.

Heavy

• Acoustic barrier wall or second-layer canopy, with an important caveat. A barrier only works where it breaks line of sight. It helps the ground-floor neighbour and does little or nothing for upper floors that see straight over it, and it is weak at low frequency. This is why a barrier can be installed, paid for, and change very little for the residents who complained.
• Isolating the plinth from the structure. Effective, expensive, disruptive.
• Relocating the DG yard, the genuine fix where the root cause is siting.
• Replacing the set. Justifiable when it is also at end of life or undersized, so the noise spend rides on a decision you were taking anyway. Do not accept a new genset as a noise remedy on its own.

On costs: we are not going to print rupee figures. The numbers circulating online are vendor quotations that vary by an order of magnitude and are not defensible. Costs here are site-specific and should be established by tender. What we can say confidently is the ordering, the free and low-cost items should always precede a capital decision.

What the rules say about recourse

[We are not lawyers, and this is not legal advice. What follows is a set of published rules, with their section numbers and dates, so that you can put them to your own advocate. Whether any of it applies to your building is a question for them, not for us.]

Be precise here, because several remedies people assume exist do not.

There are two doors, and they do different things.

The +10 dB rule is widely misunderstood in both directions. Rule 7(1) gives a person the right to complain where the noise level “exceeds the ambient noise standards by 10 dB(A) or more”, which in a residential area at night would be around 55 dB(A).

That figure is the threshold for a citizen’s complaint right under Rule 7. It is a different thing from Rule 4(1), which is written as the standard itself: “The noise levels in any area/zone shall not exceed the ambient air quality standards…”, 45 dB(A) at night in a residential area.

Two different provisions, two different numbers, doing two different jobs. Which of them bears on your building, and what follows from that, is a question for your association’s advocate. We raise it only because the two are routinely quoted as if they were the same limit.

Rule 8(1)(b) is worth reading, and is rarely quoted in this context. It empowers the authority, by written order, to prevent, prohibit, control or regulate “the carrying on in or upon, any premises of any trade, avocation or operation or process resulting in or attended with noise”, where necessary to prevent “annoyance, disturbance, discomfort or injury or risk” thereof. It does not depend on the +10 dB threshold.

Three things that do NOT apply, despite being quoted constantly:

• There is no notified in-use noise standard for DG sets. The “80 dB(A) at 1 m” figure in circulation comes from a CPCB committee recommendation circulated in draft for comment in 2014. It is not law. In one matter concerning two 2080 kVA sets, the NGT recorded a monitoring finding assessed against the 25 dB(A) insertion-loss requirement of G.S.R. 371(E) and disposed of the application, D.K. Puri v. Govt. of NCT of Delhi, O.A. No. 628/2022, order dated 19 July 2023. Note those sets were above 1000 kVA, so squarely inside the paragraph-2 regime described earlier.
• “Silence zone” is not automatic. Since S.O. 2555(E) of 10 August 2017, an area near a school or hospital does not fall into the 50/40 silence-zone category “unless notified by the State Government.” Pre-2017 guidance on this is wrong.
• The 5 dB(A) boundary rule in Rule 5(5) is written about loudspeakers, public address systems and “sound producing instruments.” Whether it reaches a generator is not something we can tell you, but note that it is frequently quoted for DG sets without anyone checking its wording.

And a nuance specific to apartment complexes: the DG yard and the complaining flats are usually inside the same premises. So the rule’s “outside his premises” obligation points at the neighbouring property, not at your own residents. For residents inside, the live hooks are Rule 4(1), which applies to any area or zone and is not bounded by property lines, Rule 8(1)(b), the builder’s obligations where the common areas have not yet been formally handed over, and common-law nuisance.

There is also an awkward structural fact worth naming: in a self-managed complex, the association is simultaneously the operator of the DG set and the body receiving the complaint about it. That is one reason an independent technical assessment is a cleaner starting point than an internal argument, it gives a committee measured facts rather than competing opinions.

Be realistic about outcomes: these routes produce directions to remediate, not damages and not an instant shutdown. CPCB’s 2014 draft guidance, the same document the “80 dB(A)” myth comes from, concedes that state boards have monitoring equipment but “scanty manpower.” For nearly every Managing Committee, the practical path is a technical audit and a remediation plan, using the legal position as leverage rather than as the primary instrument.

What your advocate will ask you for

If your committee does decide to take advice, these are the things that exist on the engineering side and take time to assemble. Gathering them is not a legal step, it is homework, and it is worth doing before the first meeting rather than after.

• Instrumented measurement by a competent party, Class 1 meter, calibrated, per the CPCB protocol, with background correction.
• Night measurements at defined positions, with date, time, duration, weather and other predominant sources logged.
• The machine’s documentation, the conformance label, rating, OEM data sheet, declared enclosure insertion loss, silencer grade, commissioning report, AMC scope, run-hour log.
• Photographic evidence of the defects, open panels, missing foam, perished seals, flattened mounts, rigid exhaust connections, corroded silencer.
• Your Consent to Operate under the Air (Prevention and Control of Pollution) Act, 1981, and its conditions. Many consents carry explicit noise and stack-height conditions, and breach of a consent condition is a far sharper lever than a general ambient complaint.
• A written record of what has already been raised internally, and with whom, your facility-management contractor, and the builder where the common areas have not yet been formally handed over.

A note on the National Building Code, and which version applies to you

The Bureau of Indian Standards withdrew the National Building Code of India 2016 on 30 April 2026 and published the National Building Construction Standards 2026 (NBCS 2026) in its place. The BIS portal record for SP 7:2016, read on 7 September 2026, states: “Withdrawn, On – 30 Apr, 2026, Superseded by IS: SP 7:2026”.

But BIS withdrawing a code does not, by itself, change what your building is held to. BIS has never had the power to make the National Building Code binding. NBCS 2026 says so about itself, in its own front matter:

“As the regulation of land development and buildings is a State subject and is dealt by the States and local bodies such as urban (and rural) local bodies within their jurisdiction through building regulatory documents like building regulations, development control rules building byelaws, and fire regulations, this document is voluntary in nature and is non-binding. The implementation depends on adoption or adaptation by concerned parties or stipulations in a contract or by appropriate adoption by the concerned authorities.”

— NBCS 2026 (SP 7:2026), front matter, page vii

A national code becomes enforceable only when a State or local authority writes it into its building bye-laws. And in Bengaluru, what the bye-laws point at is genuinely unresolved, and older than you would expect.

The Bangalore Mahanagara Palike Building Bye-Laws 2003 carry this note: “All references of clauses of the National Building Code in these bye-laws is referred to the publication of the National Building Code of India 1983. If these numbers are changed in subsequent additions, corresponding numbers shall be substituted.” (page 48; “is referred to” and “additions” are as printed). That points at the 1983 edition, with a clause that carries the reference forward as editions change.

The five city corporation amendment bye-laws gazetted on 14 May 2026, Bengaluru Central, East, North, South and West, each amend “the Bengaluru City Corporation Building Bye-Laws, 2003”, made under the Greater Bengaluru Governance Act 2024. They mention no national code at all.

The Karnataka Model Building Bye-Laws 2017 (Corporations notification) name NBC 2016 by name on 27 pages, for example, “All structural design/safety aspects as per latest BIS Codes & NBC 2016 shall be complied along with consideration of weight of Fire Engine & its manoeuvrings.”

Which of those instruments governs your building today is a question we cannot answer for you, and we are not going to pretend otherwise. The 2026 amendments are built on the 2003 bye-laws, which is a strong indication, but whether the 2017 Model Bye-Laws also apply to Bengaluru is an open question, and it is not ours to settle.

So we cite NBCS 2026, it is the current national standard and the best available statement of good engineering practice, and we name NBC 2016 where the comparison matters. What we will not do is tell you that either one is what your corporation will enforce. Put that question to your local authority, and to your association’s advocate, with the bye-law in front of you.

A note on how we quote. Bold inside a quotation is our emphasis, not the original’s. Clause text, spelling and punctuation are reproduced exactly as printed, including the occasional printing error in the source, which we do not silently correct.

Sources

• CPCB, Noise Limit for Generator Sets Run with Diesel, Schedule I, Serial No. 94, Environment (Protection) Rules, 1986, notified by G.S.R. 371(E) dated 17 May 2002, amended through G.S.R. 215(E) dated 15 March 2011. https://cpcb.gov.in/displaypdf.php?id=bm9pc2VfZGF0YS9ub2lzZWxpbWl0X2RpZXNlZ2Vuc2V0LnBkZg%3D%3D
• Ministry of Environment & Forests, The Noise Pollution (Regulation and Control) Rules, 2000, S.O. 123(E) dated 14 February 2000, as amended to S.O. 2555(E) dated 10 August 2017.
• MoEFCC, G.S.R. 804(E) dated 3 November 2022, Environment (Protection) Third Amendment Rules, 2022 (CPCB IV+ genset emission standards). Index: https://cpcb.gov.in/genset-notifications/
• CPCB, Protocol for Ambient Level Noise Monitoring, July 2015.
• CPCB, System & Procedure for Compliance with Noise Limits for Diesel Generator Sets (upto 1000 KVA), January 2008. https://cpcb.gov.in/displaypdf.php?id=Tm9pc2VfYW5kX0dlbnNldHMvZ2VuZXJhdG9yMTAwMGRpZXNlbC5wZGY%3D
• National Green Tribunal, Principal Bench, D.K. Puri v. Govt. of NCT of Delhi & Ors., O.A. No. 628/2022, order dated 19 July 2023. https://indiankanoon.org/doc/164147712/
• Supreme Court of India, Noise Pollution (V), In Re, (2005) 5 SCC 733, judgment dated 18 July 2005.
• WHO, Environmental Noise Guidelines for the European Region (2018). https://www.who.int/europe/publications/i/item/9789289053563
• CPCB, “WHO Guidelines for Noise.” https://cpcb.gov.in/who-guidelines-for-noise-quality/
• Basner, M. & McGuire, S. (2018), Int. J. Environ. Res. Public Health. https://pubmed.ncbi.nlm.nih.gov/29538344/
• IS 2974 (Part 1): 1982, Code of Practice for Design and Construction of Machine Foundations, Foundations for Reciprocating Type Machines, BIS. Named only; we do not hold this standard and quote nothing from it.
• Bureau of Indian Standards, National Building Construction Standards 2026 (NBCS 2026), SP 7:2026, Part D ‘Building Services’, Section 4 ‘Acoustics, Sound Insulation and Noise Control’: Table 4 ‘Acceptable Indoor Noise Levels for Various Buildings’, clause 4.1, printed p.18; clause 5.2.3.2 ‘Suppression of noise at the source itself’, printed p.20; List of Standards, entry (4), printed p.74.
• Bureau of Indian Standards, NBCS 2026 (SP 7:2026), Part D ‘Building Services’, Section 3, Table 1 ‘Requirements of Acoustic Comfort’ and clause 3.4.8.1, printed p.19.
• Bureau of Indian Standards, NBCS 2026 (SP 7:2026), Part D ‘Building Services’, Section 2, clause 4.3.1 ‘Location’, printed p.40; clause 4.3.3(c), printed p.41; clause 4.3.3(g) and the section’s List of Standards, entry (1).
• Bureau of Indian Standards, NBCS 2026 (SP 7:2026), Part C, Section 5, Annex E, Table 10, printed p.42 (wind sway); Section 1, Annex L, Table 50, printed p.187 (blast peak particle velocity); Section 6A, Annex C, printed p.184 (footfall). Cited to show what they do not govern.
• Bureau of Indian Standards, SP 30:2023, National Electrical Code of India (2nd Revision), a BIS Special Publication; clause 1 of Part 2/Sec 1 states that it “provides guidelines”, Part 2 Section 1, ‘Electrical Installation of Stand-by Generating Stations’: clause 1 (Scope), printed p.388; clause 5.6.1 ‘Noise pollution – Acoustic Enclosures’, printed p.392; and clauses 6.1 (siting), 6.3.2 (foundation and anti-vibration mounts) and 6.5.5 (silencer).
• IS 15575 (Part 1) : 2016 / IEC 61672-1 : 2013, Electroacoustics, Sound Level Meters Part 1 Specifications (First Revision), technical committee LTD 7, listed in NBCS 2026, Part D, Section 4, List of Standards, entry (4), printed p.74; status active, reaffirmed 2022 per the BIS catalogue record. We do not hold this standard and quote nothing from it.
• IS 13301 : 1992, Vibration Isolation for Machine Foundations, Guidelines, BIS. Named only; we do not hold this standard.
• IS/ISO 8528 (series), referred to in SP 30:2023 clause 5.6.1. Named only; we do not hold this standard.

SP 7:2026 and SP 30:2023 are BIS publications. Our copies are licensed; we quote short passages with their exact clause and printed-page coordinates so that any reader can verify them against their own copy, and we do not reproduce pages.

You May Also Like

More than a third of lift safety checks failed at handover, and residents never saw the list

Read Article →

When the Lift Fails, Who Answers? The Supreme Court Ruling Every RWA Should Read

Read Article →

Why Every Apartment Community Needs a Regular RWA Health Check-up — Before It Costs You Lakhs

Read Article →

The Hidden Crisis Threatening Your Apartment Investment: Why Every RWA Needs a Health Check Audit

Read Article →

View All Articles