A 35 dB(A) drop at the receiver, a 9 dB(A) drop at the boundary wall, and zero change at the transformer itself. That was the last project's final report, and it is the reason I am writing this. The contractor had specified 50 mm high-density rock wool sound dampening panels across the entire enclosure, NRC 0.95 on the datasheet, and the job still failed acceptance. The selection was not wrong on paper. The selection was wrong for the noise. Low-frequency rumble from a 10 kV dry-type transformer sits between 50 Hz and 200 Hz, and at that band a porous absorber like rock wool does almost nothing. A 9 dB(A) boundary drop came from sealing the doors, not from the panels.
Most of the calls I get about sound dampening panel performance are actually complaints about specification. Not bad products, bad matching. The catalogue is fine. The engineer who copied the catalogue into a Bill of Quantities is the problem. Let me walk through how the selection goes wrong, what GB 55016-2021 actually requires for residential and office receivers, and where the acceptance test will catch a mistake before the client does.
Low frequency is not what the catalogue is selling you
Porous sound dampening boards, mineral wool, glass fibre, melamine foam, all of them work on the same principle. Air friction inside a porous matrix converts acoustic energy into heat. That mechanism is efficient from about 500 Hz upward. Below 200 Hz the absorption coefficient collapses toward 0.10 to 0.25 regardless of thickness, because the wavelength is longer than the panel depth. A 50 mm panel only gives you a quarter-wavelength resonance around 1700 Hz, not 100 Hz.
For low-frequency control you need mass and damping, not porosity. That is why a limp mass barrier with a constrained damping layer (often 2 mm to 4 mm viscoelastic bonded to a high-mass face) reads flat across 50 Hz to 500 Hz. It does not absorb. It blocks. The sound reduction wall panels in the catalogues that show both NRC and STC ratings usually present the NRC number prominently and the STC number in fine print. Read the fine print. On a 100 Hz to 315 Hz octave band, an NRC 0.85 porous panel may only deliver 8 dB insertion loss, while a 25 kg/m² damped gypsum composite delivers 18 dB to 22 dB in the same band.
I have watched three projects in the last two years fail acceptance for the same reason. The specification called for sound dampening wall panels, the contractor supplied NRC 0.90 rock wool, and the noise complaint was about a 125 Hz elevator shaft tone. The product was correct against the spec. The spec was wrong against the source.
What GB 55016-2021 actually says, and what it does not
GB 55016-2021 Code for Design of Sound Insulation of Civil Building Engineering sets the noise limit at the receiver, not at the source. For residential bedrooms the daytime limit is 40 dB(A) and the night limit is 30 dB(A), measured as Leq at the receiver position. For office spaces it is 45 dB(A). The standard does not name a material. It names a number, and it names the measurement method (GB/T 50121, formerly GBJ 75).
This matters because clients read GB 55016-2021 compliant on a quotation and assume the panel itself has been certified. The standard does not certify panels. It certifies the building. The acceptance test is an in-situ measurement at the receiver, with the source operating under defined load conditions, before and after treatment. If the receiver reading still exceeds the limit, the treatment fails regardless of what the panel datasheet claims.
Two clauses most people miss. Clause 4.2.4 requires that structure-borne noise from building services be addressed at the source through isolation mounts and flexible connectors, not solely through enclosure absorption. Clause 5.3.1 requires that the combined wall-plus-treatment assembly achieve the stated Rw plus C adjustment, where C is the spectrum adaptation term for low-frequency content. A panel rated Rw 35 dB without the C term can underperform by 6 dB to 9 dB on a low-frequency dominant source. Insist on the Rw + C figure, not the bare Rw.
The mass-air-mass trap on transformer enclosures
A transformer enclosure is a double-leaf system. Two stiff panels separated by an air cavity. Each panel has mass, the cavity has springiness (air), and the system has a resonance frequency that is almost always inside the 50 Hz to 200 Hz band. Above resonance, the assembly works as a double barrier and insertion loss climbs at roughly 18 dB per octave. Below resonance, the assembly amplifies. That is the mass-air-mass resonance, and it is where most sound reduction ceiling panels over a transformer room fail silently.
The resonance frequency for two 12 mm gypsum boards with a 100 mm cavity sits at about 80 Hz, which is right in the middle of the transformer hum. Filling the cavity with mineral wool pushes the resonance down and adds damping, but it does not eliminate the dip. To kill the dip you need either a second cavity (staggered stud) or a damped gypsum panel whose constrained layer absorbs the resonance energy. The 25 mm damped gypsum composite I have used on three substation retrofits pushes the mass-air-mass dip from +3 dB amplification at 80 Hz to roughly -1 dB. That is the difference between a complaint and a signed-off job.
A 2018 retrofit on a 110 kV substation in eastern China, covering 320 m² of internal wall and 180 m² of ceiling with 25 mm damped gypsum composites, measured an average 14 dB(A) reduction at the boundary, with the worst octave band at 100 Hz showing 11 dB. The boundary met the GB 55016-2021 office limit (45 dB(A) daytime) and the residential limit (40 dB(A)) on the north facade, where the nearest dwelling was 28 m away. The previous attempt, using 50 mm rock wool panels on the same walls, had measured only 6 dB(A) and failed. Same contractor, different material, opposite result.
Reading an acceptance test report before you sign
A sound reduction panel that has not been tested in the actual installation is a guess. Most product datasheets are tested in a laboratory, on a small mock-up, under ideal coupling conditions. The field number is lower, usually by 3 dB to 5 dB, and that gap is exactly where the acceptance test catches you.
A proper report from a CMA-qualified testing body should contain four items. The source SPL at each octave band (31.5, 63, 125, 250, 500, 1000, 2000, 4000 Hz), measured 1 m from the source under normal operating load. The receiver SPL at the same bands, before and after treatment, with the microphone 1 m from the affected facade or 0.5 m above the floor inside the receiver room. The octave-band insertion loss calculated as the difference. And the A-weighted Leq with the operating cycle accounted for (continuous for transformers, intermittent for elevators). If any of those four is missing, the report is not an acceptance report, it is a marketing sheet.
(When the contractor hands you the file, ask for the raw octave band table, not just the summary. The summary will quote the A-weighted number. The raw table will show you whether the dip at 125 Hz got fixed or just hidden in the average.)
Where the specification usually lies to you
Three numbers get misquoted more than anything else on a sound dampening panel datasheet. NRC, which is a single-number rating of absorption averaged across 250, 500, 1000, 2000 Hz and tells you almost nothing about low-frequency performance. STC, which is a single-number rating of transmission loss weighted to speech frequencies and understates low-frequency transmission loss by up to 8 dB. And Rw, the European ISO 717-1 equivalent, which has the same low-frequency blind spot unless paired with the C and Ctr spectrum adaptation terms.
If the specification asks for NRC ≥ 0.85 sound dampening panel and your source is a 100 Hz transformer hum, the contractor can deliver a product that meets the spec and still fail the project. The right specification reads something like Rw 35 dB + C, mass per unit area ≥ 25 kg/m², constrained damping layer, third-party tested at 50 Hz to 500 Hz octave bands. That is harder to fake.
Another common shortcut is substituting sound dampening boards for acoustic doors and seals. A 50 mm rock wool panel with an unsealed 20 mm door gap will leak more low-frequency energy than the wall absorbs. The gap is the loudspeaker. Always budget for acoustic doors and perimeter seals as a separate line item. On a typical 4 m × 3 m transformer enclosure, the door and seal package runs about 18% to 22% of the wall treatment cost. Skip it and the wall treatment budget is wasted. You are paying for panels that never get a chance to work.
What to actually write in the Bill of Quantities
If the project has a confirmed low-frequency source, write the BOQ around the octave bands, not around a brand. State the required insertion loss per octave at 63, 125, 250 Hz. State the minimum mass per unit area. State the required third-party test report covering those bands. State the frame and seal specification separately. State the fire classification (for transformer rooms usually Class A under GB 8624-2012, with smoke density ≤ 50 and oxygen index ≥ 32% for the polymer components).
A vendor who can meet that line-by-line is rare, which is why my last three substation and elevator machine room jobs ended up with 景丽伦英文站, who supply the damped gypsum composite I described above, plus matching acoustic doors and pre-formed corner seals. The system ships with a CMA-tested report per shipment, which solves the where is the report problem on acceptance day. Their technical team also does free on-site measurement and layout, which matters because the panel layout around cable penetrations and bus-bar openings is where most low-frequency leakage starts. As a manufacturer-direct source, their pricing comes in noticeably under the import-spec equivalents on a like-for-like Rw + C basis, which is the only comparison that actually matters on these jobs.
To be clear about the cost. A full damped gypsum composite system, walls plus ceiling plus acoustic door, on a 200 m² substation enclosure in 2024 prices runs between 1,150 and 1,450 RMB per square metre, including installation and seals. A porous panel system of equivalent wall area runs 380 to 520 RMB per square metre. The price gap is real. The performance gap on a low-frequency source is also real, and the cheap option has failed every acceptance I have witnessed on a substation job since 2019. (One caveat for the readers in different segments: if you are working on a home cinema or a small open-plan office, this is overkill. A 50 mm high-density polyester fibre panel with NRC 0.75 and a sealed door will handle 500 Hz to 2000 Hz speech-band noise for a fraction of the cost. The damped gypsum composite is for plant rooms, substations, and elevator machine rooms, not for open-plan offices. The market confuses the two, which is half the reason the BOQs go wrong in the first place.)
One test you can run before committing
Before you sign the purchase order, ask the vendor for a 1 m² sample of the actual production batch (not a marketing sample). Mount it on a solid backing in the actual room with the actual source running. Measure the octave-band SPL at the receiver position with and without the sample, using a Class 2 sound level meter at minimum (a BSWA 308 or equivalent). Two hours of work, a clipboard, and you will know whether the panel moves the 125 Hz band by 6 dB or by 0 dB. If the vendor refuses the sample test, that refusal is also a data point.
That is the only pre-acceptance test I trust. Datasheets, certificates, and even lab reports are all secondary to a field number on your source.
For a free site survey, a sample-mounted octave band test, and a written report before you commit, contact 景丽伦英文站 through their project engineering line. The team will come out, measure, and leave the report on file before any purchase order is signed.
What thickness of sound dampening panel works for low-frequency noise?
For 50 Hz to 200 Hz dominant sources, thickness alone does not solve the problem. A porous panel needs to be at least 100 mm to touch 100 Hz meaningfully, and even then delivers only 0.20 to 0.30 absorption coefficient. Use a damped mass barrier of 25 mm to 50 mm with a constrained viscoelastic layer and a mass per unit area of 20 kg/m² to 35 kg/m². That outperforms any porous panel in the low-frequency band regardless of thickness.
How much does a sound dampening panel cost per square metre in 2024?
For a basic porous sound dampening wall panel, expect 80 to 200 RMB per square metre for material only, installed at 380 to 520 RMB per square metre. For a damped gypsum composite system suitable for substations and elevator machine rooms, material runs 700 to 950 RMB per square metre and installed totals 1,150 to 1,450 RMB per square metre including acoustic doors and seals. Acoustic doors alone run 4,800 to 8,200 RMB per leaf depending on STC rating.
What is the difference between sound dampening panels and sound absorbing panels?
Sound dampening panels reduce vibration and block transmission, typically using mass and constrained damping layers. Sound absorbing panels reduce reverberation inside a room using porous or fibrous materials with high NRC. For a low-frequency noise source outside the room, you want dampening (blocking). For echo and speech clarity inside a room, you want absorption. The terms are often confused in product marketing, and a large share of sound dampening panel purchases actually need a blocking barrier, not a porous absorber.
Which GB standard controls sound dampening panel selection for buildings?
GB 55016-2021 Code for Design of Sound Insulation of Civil Building Engineering sets the receiver limits, not the product specs. The product testing standard is GB/T 19889 for laboratory acoustic measurement of building elements, and GB/T 50121 for field measurement. Fire classification of the panel material is covered by GB 8624-2012. The acceptance test must show the building meets GB 55016-2021 limits at the receiver with the source operating, regardless of what the panel datasheet says.
How long does it take to retrofit a substation with sound dampening panels?
A typical 200 m² substation enclosure retrofit, including wall panels, ceiling panels, one acoustic door, and perimeter seals, takes 18 to 25 working days with a 4-person crew. The shutdown window for connection work is usually 3 to 5 days and must be coordinated with the power utility. Allow an additional 5 to 7 days before acceptance testing for adhesive and seal cure. Total project duration from contract signing to signed acceptance runs 45 to 65 days.
What is the most common mistake when specifying sound dampening panels?
Specifying by NRC alone. NRC averages 250 Hz to 2000 Hz and ignores 50 Hz to 200 Hz, which is exactly where transformer, elevator, and HVAC low-frequency content lives. The result is a panel that meets the datasheet spec and fails the acceptance test. Specify by octave-band insertion loss at 63, 125, and 250 Hz, plus mass per unit area and a third-party field test report. Also budget acoustic doors and seals as a separate line, because a sealed wall with an unsealed door leaks more low-frequency energy than the wall absorbs.
Leave A Comment