Why Duct-Borne Noise and Duct Breakout Are Different Faults
Why this matters
A silencer goes in at the fan discharge, the submittal shows 25 dB of insertion loss, and the conference room still rumbles. Everybody assumes the silencer underperformed. It did not. The noise reaching that room never went through the silencer, because it left the duct through the sheet metal above the ceiling rather than through the diffuser. Duct-borne noise and duct breakout are two different faults with two different treatment locations, and the fastest way to waste a mechanical-room mobilisation is to treat one while the complaint is the other. Telling them apart takes two microphone positions and one blanked diffuser.
The two paths, defined by where the energy leaves
Duct-borne noise travels in the airstream, from the fan, through every fitting, and out at a terminal device: a diffuser, a grille, a register, a louvre. Everything between the fan and that terminal attenuates it - duct length, lining, bends, plenums, branch splits, the terminal unit itself - and a silencer in that path is one more attenuating element. It also picks up energy on the way, because a damper or a takeoff near the terminal regenerates its own noise downstream of everything upstream of it.
Breakout is the noise radiating outward through the duct wall along the run. It leaves the system wherever the duct happens to be, which on a supply main is very often above the ceiling of a room that is not the room the duct serves. Nothing downstream of that point attenuates it, and nothing upstream of it that is downstream of the fan attenuates it either, except whatever the energy passed through before it got there.
That single sentence is the whole card: breakout is attenuated by the duct wall and by the ceiling under it, and by nothing else.
The field key that separates them
Three checks, in this order, all with the fan in its normal operating state.
Where in the room is it loudest. Duct-borne noise peaks under the terminal and falls off as you move away. Breakout peaks under the duct run, which is usually somewhere else entirely. If the mid-room level under a duct main is higher than the level under a diffuser, in the same band and on the same time basis, you are not looking at a terminal problem.
Blank the terminal. Seal the diffuser face with a rigid blank and gasket, run the system, and re-measure at the same position. Duct-borne noise collapses. Breakout barely moves. This is the decisive test and it takes minutes. Do it on one terminal at a time and put the system back before you leave: blanking a terminal changes the system's resistance, and blanking several on a system with a variable-speed fan or a pressure-independent box changes the operating point rather than only the sound.
Look at the shape of the spectrum. Breakout is dominated by low frequencies. Duct-borne noise reaching a room through a lined path and a terminal has usually lost most of its high-frequency content already, so it is also low-frequency weighted, but it will carry any regenerated content from the last fitting before the terminal, which breakout cannot. A hiss at a diffuser is never breakout.
Why the spectra differ, and why duct shape decides so much
The duct wall is a panel, and a panel's transmission loss is mass-controlled over most of the range that matters here: increasing surface mass raises transmission loss, at a stated frequency and for a stated angle of incidence, by roughly 5 to 6 dB per doubling of mass. Both parts of that condition matter, because the same doubling buys nothing where the panel is stiffness-controlled or near a resonance, and a duct wall is not an idealised limp panel.
The shape effect is larger than the mass effect and it is the one worth acting on. A rectangular duct wall is a flat plate that flexes easily, so its low-frequency breakout transmission loss is poor. A round duct is stiff in the hoop direction and radiates far less at low frequency; a spiral seam adds more stiffness again. Flat oval sits between them. The published breakout transmission loss for a given size and gauge belongs to the duct manufacturer's data or to the tested figures in the handbook chapter, per octave band, and it is not a single number.
This is why the same air volume, at the same velocity, in the same ceiling cavity, can be a complaint as a large rectangular main and a non-event as round spiral. It is also why breakout is worst close to the fan, where the internal sound pressure is highest and the duct is largest, and why a long unlined rectangular main crossing a quiet space is the classic breakout geometry.
The location rule for a silencer
A silencer only attenuates the part of the path downstream of it. Insertion loss is published for the duct-borne path, measured to a test method such as ASTM E477, and it says nothing about what leaves the duct wall on either side of the unit. The silencer's own casing has a breakout transmission loss, published separately, and on a unit sitting in a mechanical room next to an occupied space that figure can be the governing one.
So placement follows the complaint, not the equipment. Where the complaint is at a terminal, the silencer belongs upstream of that terminal and downstream of any regenerating fitting. Where the complaint is breakout above a room, a silencer at the fan is spending money on a path the complaint never used.
Worked case: a conference room rumble
A conference room shares a wall with a mechanical room, and a large rectangular supply main crosses above its lay-in ceiling on the way to the far side of the floor. The room's own diffusers are fed by a branch off that main. All levels below are unweighted octave-band sound pressure levels re 20 micropascals, 1-minute equivalent-continuous, microphone at 1.2 m above the floor.
Position A, directly under a diffuser, fan running: 63 Hz band, 54 dB. Position B, mid-room under the main, away from any diffuser, fan running: 63 Hz band, 58 dB. Position B, fan off, nothing else changed: 63 Hz band, 40 dB.
The margin at position B is 18 dB, which is at or above 10, so the combined reading is the source's own level within a few tenths and the background correction was declined and recorded as declined. Position B reads 4 dB higher than position A while sitting further from every terminal in the room, which already points away from the airstream.
The blank test. A gasketed blank over the diffuser at position A dropped the 63 Hz band there by 1 dB, from 54 to 53, and left position B unchanged at 58 dB. Duct-borne noise through that diffuser is not the complaint. The blank came off before the crew left the room.
What got done. The main over the conference room was replaced with round spiral of equivalent free area for the length of the crossing, and the two branch takeoffs in that length were relocated. No silencer was fitted and nothing was done in the mechanical room.
Result, same position B, same basis: 63 Hz band, 44 dB, against the same 40 dB background. That margin is 4 dB, which sits in the 3-to-10 band where a correction applies, so 2.2 dB comes off the combined reading and the source contribution is reported as 41.8 dB in the 63 Hz octave, unweighted, re 20 micropascals, at that position. Not "the room is at 44", which would be reporting the background as if it were the fan.
Sibling-rule check. Every level carries quantity, weighting, bandwidth, time basis, reference and position, and the before and after figures share all six. The background margin gate is applied per reported figure using the three zones the sibling procedure states, and the correction applied at a 4 dB margin is the one that table gives rather than a subtraction of the raw background. The corrected after figure is compared against a before figure taken against the same background at the same position, so no corrected number is benchmarked against an uncorrected one. The mass-law relationship is stated with its frequency and incidence conditions, and the shape effect is routed to published per-band data rather than given as a single number.
Break-in, which is neither of the above
The reverse case has its own name and gets missed because nobody looks for it. A duct passing through a noisy space - a mechanical room, a kitchen, a loading area - picks up airborne noise through its wall, and that energy then travels inside the duct and out at a terminal in a quiet space some distance away. The symptom is a complaint at a diffuser that changes when a door in a completely different room opens, and the treatment is at the duct where it passes through the noisy space, not anywhere near the complaint. The same wall properties govern, in the other direction.
Working above a ceiling to run these checks
Access is the hazard, not the sheet metal. Where the work is off a ladder, set and secure it under 29 CFR 1910.23 and keep both feet and one hand on it while reaching, and where it is off a platform or near an opening, fall protection under 29 CFR 1910.28 applies. Cut edges on sheet metal cause deep hand lacerations, so handle open duct ends with cut-resistant gloves. Do not disturb any existing duct insulation or lining you cannot identify, because thermal system insulation in older buildings may be asbestos-containing and that determination is made under the building's own asbestos program and 29 CFR 1910.1001 before any material is cut; where the lining is identified fibrous glass, cutting it releases airborne fibre, so use the respiratory control the product's safety data sheet specifies under a 29 CFR 1910.134 program rather than a nuisance dust mask.
How to verify you fixed the right fault
Re-run the two positions and the blank test rather than only the position that was complained about. The confirming pattern for a breakout fix is that position B falls and position A does not, because you did nothing to the airstream. If both fell, something else changed at the same time and you do not yet know what you bought. If position B fell and the occupant still complains, re-measure with the diffuser blanked, since a breakout fix that reveals a previously masked terminal noise is the normal second act on this fault and it is a different job with a different treatment.
References
- ASHRAE Handbook chapters on sound and vibration control, for duct breakout transmission loss data by shape, size and gauge, and for the conditions those figures were derived under
- ASTM E477, the laboratory method for silencer insertion loss and pressure drop, in the edition the tested product's report names
- Duct and silencer manufacturer's published breakout transmission loss and insertion loss for the specific product
- 29 CFR 1910.23 and 29 CFR 1910.28, ladders and fall protection, for ceiling-cavity access
- See related: What a Duct Silencer Costs You in Pressure; Why Fan Noise Is a Selection Problem Before a Treatment Problem