A tool that predicts the insertion loss of a lined HVAC duct silencer using the Sabine empirical formula. Adjust duct width, height, silencer length, absorption coefficient and airflow velocity to see at a glance the trade-off between frequency-dependent attenuation, flow-generated noise and pressure drop.
Parameters
Duct width W
mm
Duct height H
mm
Silencer length L
m
Length of the lined section along the flow direction
Absorption coefficient α
Mid-band representative value; ~0.8 for 50 mm glass wool
Airflow velocity v
m/s
Drives flow-generated noise and pressure drop
Evaluation frequency f
Hz
Frequency at which the central IL is reported
Results
—
Hydraulic diameter d_h (m)
—
Sabine attenuation (dB)
—
Frequency factor K_f
—
Insertion loss IL (dB)
—
Flow-generated noise L_w (dB)
—
Pressure drop ΔP (Pa)
—
Duct cross-section and wave attenuation — visualisation
The vertical rectangle is the duct cross-section; hatched bands on top and bottom are the absorptive lining. The internal wave shows how the incoming sound decays in amplitude (red → blue) as it passes through the silencer. Arrows below indicate airflow direction.
Insertion loss IL (dB) and hydraulic diameter d_h. P is the perimeter (m), A is the cross-section area (m²), L is the silencer length (m), α is the absorption coefficient, K_f is the 63-8000 Hz octave-band correction factor.
$$L_w \approx 10\,\log_{10}\!\bigl(v^{6}\bigr) + 10,\qquad \Delta P \approx \tfrac{1}{2}\,\rho\,v^{2}\cdot\zeta\cdot\tfrac{L}{d_h}$$
Flow-generated noise L_w (dB, rough) and pressure drop ΔP (Pa). ρ is air density (1.2 kg/m³), v is airflow velocity (m/s), ζ is a silencer shape factor (0.05 here). The v^6 scaling is the standard HVAC empirical law.
Attenuation characteristics of duct silencers
🙋
Why are the big HVAC ducts you see above office or school ceilings so quiet? The motor must be loud, yet only a soft "swoosh" comes out of the diffusers.
🎓
Inside there is a "duct silencer". The construction is simple: about 50 mm of glass wool or rock wool lining is glued to the inner walls of the duct. Air still flows through normally, but the sound wave is converted to heat by friction inside the fibers, and the amplitude decays as it travels. Predicting how much it decays with Sabine's empirical formula IL = 1.05·α^1.4·(P/A)·L is the starting point of HVAC acoustic design.
🙋
I made the duct width W narrower and the IL jumped up. Why? I thought only the area should matter.
🎓
Good catch. The key term in the formula is P/A — perimeter divided by area. For the same area, a flatter or longer-and-thinner duct has more perimeter, so the sound touches more lining per unit volume. For instance, a 400×400 mm duct gives P/A = 1.6/0.16 = 10; a 200×800 mm duct of the same area gives 2.0/0.16 = 12.5, about 1.25× more attenuation. As long as space allows, "flat and long" is acoustically the better cross-section.
🙋
If I set the frequency to 100 Hz or 6000 Hz, the IL drops a lot. So the silencer is bad at both low and high pitches?
🎓
Yes — a lined silencer is good in the mid-band (500-2000 Hz) and weaker at the extremes. Two reasons. First, the absorber peaks where its thickness is about a quarter of the wavelength (the λ/4 rule); 50 mm wool gives a peak near 1700 Hz. Second, at very low frequencies the wavelength is larger than the duct cross-section, so the plane wave slips through with little interaction. For low-frequency attack you usually combine a Helmholtz-style reactive silencer or a thicker absorber.
🙋
Push the velocity to 15 m/s and the flow-generated noise tops 60 dB. We silence the source and then create a new noise of our own?
🎓
That's the silencer-design pitfall. The turbulent boundary layer over splitters and lining radiates noise whose acoustic power roughly scales with v^6. Going from 10 to 15 m/s multiplies the power by 7.5× (about 8.8 dB). So "duct is noisy → add a longer silencer → cross-section is unchanged → velocity rises → flow noise grows → no net gain". Fixes are (1) enlarge the duct in the silencer section to drop velocity below 10 m/s, (2) streamline splitter leading/trailing edges, or (3) use a low-noise splitter with a perforated facing.
🙋
How do you balance pressure drop against insertion loss in practice?
🎓
For general HVAC, aim for ΔP = 30-100 Pa and IL = 10-25 dB, keeping the silencer ΔP under 10% of the fan's total static pressure. For data centers or operating theaters you may want IL ≥ 30 dB and tolerate ΔP near 200 Pa, sizing the fan accordingly. For low-pressure systems like residential ventilation (≈50 Pa total), ΔP must stay under 10 Pa, so a short, wide "plenum-style" silencer is preferred. Never chase insertion loss alone — solve it together with ΔP, flow rate and fan capability.
Frequently Asked Questions
The most widely used empirical relation is Sabine's (1948) formula: IL = 1.05 · α^1.4 · (P/A) · L, where α is the absorption coefficient of the wall lining, P is the cross-section perimeter, A is the cross-section area, and L is the silencer length. The form shows that a duct with a large perimeter-to-area ratio (long-and-thin or flat ducts) exposes more lined surface per unit volume and therefore attenuates more. This tool multiplies the Sabine value by an octave-band correction factor K_f (63-8000 Hz) to give the band-dependent insertion loss IL.
A typical duct silencer with about 50 mm of glass-wool lining reaches its peak insertion loss in the mid-band, 500-2000 Hz (K_f ≈ 1.0). This is because the absorption coefficient of porous materials peaks near the quarter-wavelength frequency (λ/4 rule): a 50 mm lining peaks around 1700 Hz. At low frequencies (below 250 Hz), K_f drops to about 0.5 and a single silencer cannot suppress transformer hum or large-fan blade-pass tones; at high frequencies (above 4000 Hz), K_f falls to 0.6-0.8 because viscous losses inside the fibers diminish. Add a perforated facing or use a thicker lining to extend the high-frequency response.
The silencer's own attenuation barely depends on velocity, but high-speed airflow over splitters and lining surfaces generates self-generated (flow regenerated) noise, which appears downstream of the silencer as a new source. As a rule of thumb the radiated sound power scales with v^6, and this tool estimates L_w ≈ 10·log10(v^6) + 10 dB. In practice, keep the silencer-section velocity below 10-15 m/s; if you need more flow, enlarge the cross-section to slow the air, or pick a low-noise splitter silencer.
For general HVAC use, a single silencer typically loses ΔP = 30-100 Pa, and the rule of thumb is to keep the silencer's ΔP under 10% of the fan's total static-pressure budget. Pressure drop scales roughly as ΔP = ζ · (1/2) · ρ · v², where ζ is the shape factor of the lining intrusion (0.03-0.10). This tool uses ζ ≈ 0.05 · (L/d_h), so a longer silencer attenuates more but also drops more pressure. Always co-optimise insertion loss, pressure drop, fan size and electrical consumption.
Real-World Applications
HVAC design for offices and commercial buildings: Large ducts running above office ceilings always have a lined silencer right after the air-handling unit (AHU) outlet and again before each floor branch. ASHRAE and JIS guides specify a background level of NC-35 (~40 dBA) for office spaces; reaching it typically requires 25-35 dB of attenuation from the AHU outlet's sound power. A Sabine-style estimate fixes the silencer length (1.2-2.4 m is typical) and cross-section ratio first, then the design is fine-tuned with the vendor's measured data.
Data-center noise control: In data centers, large air-handlers run 24/7 to cool dense racks, and meeting an outdoor noise limit (e.g. 50 dBA in a residential zone) makes duct silencers especially critical. Because cooling demand requires high airflow, exceeding 15 m/s in the silencer section leaks flow-generated noise outdoors. Multi-stage silencers combined with reactive (resonator) elements are common, and a Sabine-style estimate sets the overall acoustic budget before detailed CFD-aero-acoustic design.
Hospitals, recording studios, and concert halls: Operating theaters and ICUs target NC-25-30 (≤ 35 dBA); recording studios and concert halls aim at NC-15-20 (25-30 dBA). The demand on duct silencers becomes extreme. To kill the low-frequency floor noise, reactive (resonator) silencers and lined silencers are stacked in series, and the total assembly can reach 3-5 m. Section velocities are also held below 5 m/s so that flow-generated noise stays under 25 dB.
Residential ventilation and kitchen exhaust: Heat-recovery ventilators and balanced ventilation systems also need compact silencers. Residential systems have a very low static-pressure budget, so a short box-style (plenum) silencer targeting IL = 10-15 dB is preferred. Kitchen exhausts clog glass wool with grease, so washable perforated-metal silencers backed by ceramic fiber, or pure reactive silencers, are the standard solution.
Common Misconceptions and Pitfalls
The most common pitfall is assuming that the Sabine IL is exactly what you will measure on site. The Sabine equation is derived assuming (1) no flow, (2) plane-wave incidence and (3) no reflected waves. Real ducts always have elbows, branches and area transitions that reflect waves back, and at low frequencies the on-site IL often falls to 50-70% of the calculated value. In addition, if you cannot afford a straight run of at least three duct diameters either side of the silencer, distorted inflow adds further loss. Practical design uses 0.7-0.8× the Sabine value as the expected IL and then confirms with the manufacturer's measured spec.
Next, treating the absorption coefficient α = 1 as physically reachable. α = 1 is the limit of perfect absorption; even 100 mm glass wool tops out near α ≈ 0.95 at 1000 Hz. α is also strongly frequency-dependent, so using the "average α" from a catalog directly in IL calculations underestimates the low-frequency response. Always pick the α value at the design frequency (commonly 500 Hz and 1000 Hz) and remember that this tool's K_f reproduces the frequency dependence by adjusting the Sabine value. Note also that α drops 10-20% over 5-10 years due to dust accumulation and moisture uptake, so maintenance planning matters.
Finally, thinking that the duct silencer alone completes the noise control. The fan radiates noise not only via the ducted path but also via casing vibration. Even 30 dB of duct attenuation does nothing if casing vibration travels through walls and floors. Real designs combine (1) vibration isolators under the fan, (2) flexible duct connectors, (3) duct silencers, and (4) room-side absorptive finishes to actually meet the target NC. The IL value from this tool is "acoustic power attenuation along the duct path" only — it is not a guaranteed in-room noise level.
How to Use
Enter rectangular duct dimensions: width (mm) and height (mm). Standard commercial HVAC ducts range 200×100 mm to 1000×500 mm.
Set silencer length (mm). Typical installations use 600–1200 mm lined sections for 500–2000 Hz attenuation.
Input absorption coefficient (0–1.0) for lining material. Fiberglass board yields 0.6–0.8 at mid-frequencies; melamine foam achieves 0.5–0.7.
Review hydraulic diameter, Sabine attenuation in dB, frequency factor, insertion loss, flow noise, and pressure drop across the silencer.
Worked Example
A rectangular duct 400 mm wide × 250 mm high with a 900 mm fiberglass-lined silencer section (α = 0.70) produces: hydraulic diameter d_h = 0.320 m; Sabine attenuation ≈ 12.5 dB at 1 kHz; frequency factor K_f = 1.0; insertion loss IL ≈ 11.8 dB; flow-generated noise L_w ≈ 78 dB (at 4 m/s velocity); pressure drop ΔP ≈ 18 Pa. This configuration reduces downstream VAV box noise by 12 dB, meeting ASHRAE 90.1 requirements for office spaces (35 dB NC criterion).
Practical Notes
Attenuation improves with longer silencers and higher absorption coefficients, but pressure drop increases roughly as length². Balance noise control against fan energy penalties.
Frequency factor K_f accounts for cross-sectional geometry: smaller hydraulic diameter yields greater high-frequency attenuation. Aspect ratios above 4:1 reduce effectiveness.
Flow-generated noise dominates in supply ducts above 5 m/s; silencers become critical for noise attenuation. Verify insertion loss exceeds 10 dB for VAV terminal boxes in occupied zones.