The Partition That Was Built Right and Still Failed

Why this matters

The hardest sound complaint to work is the one where every trade did what it was told. The wall type matched the schedule, the submittal was approved, the sealant was there, the fire inspection passed, and the tenant can still follow a conversation through the wall. When that happens the instinct is to look for a workmanship failure, and on this job there was not one. The failure was in a joint that had been qualified for a fire test and was then assumed to be qualified for an acoustic one, and the whole case was decided by a single correctly answered request for information months before anyone held a meter. This one is worked backwards from the paperwork, because that is where it was solvable.

The complaint, and what was already in the file

A demising wall between two suites in a multi-tenant medical office building, 40 ft long and 9 ft floor to deck, 360 ft2 of common area. The complaint was speech intelligibility: not a hum, not a bass line, words.

What the file held before anyone went to site:

• Specification. STC 55 minimum at demising walls, laboratory rating to ASTM E90 classified per ASTM E413, in the editions the specification named.
• Approved submittal. A tested assembly, report number referenced, deck to deck, cavity insulation, resilient channel one side, acoustical sealant at head and base both faces.
• Wall type schedule. The demising line correctly tagged with that wall type over its whole length.
• Firestop submittal. A head-of-wall joint system listed for the fire-resistance rating, compressed mineral wool safing in the flutes with a joint spray over it, approved by the fire-protection subcontractor and by the authority having jurisdiction.
• Punch photographs. The head of wall, photographed for the fire inspection.
• Closeout. Fire inspection passed. No acoustic verification anywhere in the contract.

A file that clean is itself a finding. Every requirement in it was met.

Reading it in reverse

Start from the outcome and work back to the last document that could have prevented it.

The field test came first, because a complaint without a measurement is an opinion. Measured to ASTM E336 and classified on the same E413 basis, the wall returned an apparent sound transmission class of 44 against a laboratory 55. Apparent is the operative word and a sibling card owns it: the field figure includes every path, and it is a different quantity from the laboratory rating, not a degraded version of it.

Eleven points is far too large to be workmanship on a wall that photographs correctly. So the question became: where can 11 points hide in an assembly nobody built wrong? There are only three answers. A path that goes around the wall. An element in the wall that is not wall. Or an aperture. The suites shared a topped slab, the wall ran deck to deck so the plenum was not continuous over it, and the wall carried no door and no duct crossing. That left an aperture.

The arithmetic that named the size of it

Transmission adds on area-weighted transmission coefficient, not on decibels, which is the small-hole card's rule and the reason this next step works at all.

• Field: 44 on the E413 basis, so the coefficient is 10^-4.4 = 3.98e-5. Times 360 ft2, the whole assembly is passing the equivalent of 0.0143 ft2 of open area.
• The wall itself at its laboratory 55: 10^-5.5 = 3.16e-6. Times 360 ft2, that is 0.0011 ft2.
• The difference is 0.0132 ft2, which is 1.9 square inches.

Eleven points of a 360 square foot wall, in a medical building, on a suite that photographs perfectly, is the equivalent of a hole smaller than a business card. That number is what got the building owner to fund an investigation.

Written as a bound, not a value. This arithmetic charges the entire excess to airborne leakage in the plane of the wall. If any part of it is structure-borne flanking, the leak is smaller. So the honest statement is that the equivalent open area is at most 1.9 in2, written with one inequality sign, and it is treated as an upper bound throughout.

Correction, and what it re-bases. The 0.0011 ft2 charged to the wall is derived from a laboratory rating, which already contains a sealed test-frame perimeter and no flanking. Using it as the wall's field contribution therefore understates the wall and overstates the leak, which is the conservative direction for a bound and the wrong direction for an apportionment. It is used here only to size the aperture, never to credit the wall.

Ceiling check, printed. An aperture of 0.0132 ft2 in 360 ft2 is an area fraction of 3.7e-5, which caps the composite at 10 log10 (1 / 3.7e-5) = 44.4 dB no matter how good the rest of the wall becomes. Measured 44. The two agree, which is the confirmation that the aperture model fits and that no fourth layer of board was ever going to help.

The request for information that decided it

Working back through the log, one entry stood out. During framing, the drywall subcontractor asked whether acoustical sealant was required at the head of wall where the rated joint system occurred. The answer came back: provide per the rated assembly listing.

That answer was correct. A listed fire-resistance-rated joint system is qualified as an assembly, tested to ASTM E1966 or UL 2079 in the edition the adopted code references, and what may be installed in that joint is fixed by the listing. Adding a material the listing does not include voids it. The firestop contractor read the answer as an instruction not to add anything, which is what it said.

What nobody wrote down is that a joint qualified for fire is not thereby qualified for air. The listing's performance criteria are about fire endurance and hose stream, not about airtightness, and compressed mineral wool is porous by design. The joint did exactly what it was listed to do, over 40 lineal feet, at the top of the wall, above a ceiling where nobody was ever going to look at it acoustically.

Listed head-of-wall systems exist that carry both a fire rating and a tested acoustic rating in the same listing. Specifying one of those is the fix, and it is a submittal decision, not a field decision.

Confirming it without cutting anything

The confirmation was visual and differential, and nothing rated was touched.

Before going above the ceiling: set the ladder to the rules that govern your work, 29 CFR 1910 Subpart D in general industry or 1926 Subpart X for construction, and do not disturb thermal system insulation or surfacing material of unknown vintage, which is presumed to contain asbestos until sampled, with non-disturbance and sampling as the control under 29 CFR 1910.1001 and 29 CFR 1926.1101 rather than a dust mask.

Do not cut, compress, remove, probe into or add material to a fire-rated joint system. Any of those voids the listing and leaves a rated assembly out of compliance until it is restored by that system's own listed method. Inspection above the ceiling is visual only, and anything opened in a rated assembly gets restored before the crew leaves the floor. Acoustical sealant is not a firestop material and a rated joint is not a place to try something.

So the differential test moved the source instead of closing the path. Nothing was sealed and nothing was touched. Place equipment in a plenum clear of sprinkler piping and heads and do not lean anything against them, because sprinkler obstruction and pipe support are governed by NFPA 13 in the edition the authority having jurisdiction adopted, which binds the owner and reaches the contractor through the permit.

• Loudspeaker in the source suite's room, level referenced to the source-room average: the receiving-room level gave the room-to-room result already measured.
• Loudspeaker moved into the source suite's above-ceiling space, level referenced to the source-plenum average: the receiving-side plenum ran 9 dB higher relative to its own source level than the room-to-room path did. Air above the ceiling was the strong route.
• One ceiling tile lifted on the receiving side and a scan taken along the head of wall, 500 Hz octave band, unweighted, re 20 micropascals, 30-second equivalent-continuous: the level peaked at each deck flute and fell about 7 dB between flutes. The flutes are exactly where the safing sits.
• Re-caulking the base of the wall on both faces returned nothing measurable.

The head condition owned the complaint, and the evidence is spatial rather than inferential.

Sibling-rule check, printed. Composite transmission summed on area-weighted coefficient rather than by subtracting decibels: yes. The field figure named as apparent transmission class and never compared directly to the laboratory number without saying they are different quantities: yes. The single-path gain quoted against the aperture ceiling that caps it: yes. The leak written as a one-sided bound, since the character of the residual is unknown: yes. No rounding taken in the direction that flatters the wall: the laboratory coefficient used for the wall pushes the bound higher, and that is stated.

What the record never had a place for

Three gaps, and they are the transferable part of this job.

No acoustic verification requirement existed. The contract required a laboratory-rated assembly and an installed condition. It never required a field measurement, so the first measurement in the building's life was taken by a consultant hired after a tenant complained. A field test at one representative wall during construction would have caught this while the ceiling was still open.

No document assigned the head condition an acoustic performance. The wall schedule carried the rating. The firestop submittal carried the fire rating. Nothing carried the sentence that both had to be satisfied by the same listed system.

The photographs answered the wrong question. The head of wall was photographed for a fire inspection, so the images prove the safing was installed and prove nothing about continuity of an air seal. A photograph is evidence of whatever the person taking it was looking for.

What the shop can bank operationally: when an assembly has to satisfy two different ratings at the same joint, ask for one listing that carries both, in writing, at submittal. Whether a correctly answered request for information shifts any liability is a question for the reader's own attorney; that the joint condition was decided at submittal and not in the field is an operational fact you can act on.

References

• ASTM E90, ASTM E336 and ASTM E413 for laboratory and field airborne sound performance, and ASTM E1966 or UL 2079 for fire-resistance-rated joint systems, each in the edition the specification or the adopted code cites; these are consensus standards and bind through that reference
• NFPA 13, in the edition adopted by the authority having jurisdiction, for sprinkler obstruction and coverage
• 29 CFR 1910 Subpart D and 29 CFR 1926 Subpart X for ladders; 29 CFR 1910.1001 and 29 CFR 1926.1101 for presumed asbestos-containing materials
• See related: What a Small Hole Does to a Large Wall; Why a Wall Never Performs Like Its Rating; How to Firestop a Penetration So the Rating Survives; What a Fire-Rated Assembly Is and How It Gets Broken