Frost on the Roof Deck: Air Leakage or Vapor Diffusion?
A six-story, 96,000 ft2 mixed-use building in a cold climate (ASHRAE CZ 6A) has a fluid-applied vapor-permeable air barrier over exterior gypsum sheathing, and a 6-mil polyethylene vapor retarder on the warm side of the stud cavity. During a winter walkthrough at outdoor -8 C (18 F) and indoor 22 C (72 F) with 38% RH, the auditor finds frost and dark staining on the underside of the steel roof deck above the top-floor corridor, and damp gypsum at the wall-roof intersection. A whole-building test measured 2.9 ACH50 against a 2.0 ACH50 program limit, with the largest leakage indicated at the top-floor parapet line. The contractor proposes adding a second interior vapor retarder coating and increasing roof insulation. No interior finishes are installed yet.
1. Identify the most probable failure mechanism and explain why the contractor's proposed remedy would not resolve it.
Read the worked answer
The staining and frost pattern point to air exfiltration, not diffusion. At 22 C and 38% RH the dew point is about 7 C, so any surface below roughly 7 C will condense or frost when humid indoor air reaches it. The underside of a steel deck in winter is far below that, and the deck is directly coupled to the parapet where the whole-building test located its largest leakage. Concentrated frost on one localized detail is the signature of a discontinuous air barrier plane, since diffusion moves moisture slowly and uniformly, not in a plume at a joint.
The proposal treats the wrong control layer. Extra interior retarder does not stop air movement, and a second retarder in a CZ 6A wall further restricts inward drying. Added roof insulation warms the deck somewhat but does not interrupt the exfiltration path. The auditor should require the parapet and wall-roof transition to be opened, the air barrier made continuous and terminated to the deck, and the assembly retested before any insulation or finishes are added.
2. The gap between measured and target leakage is 0.9 ACH50. What does that mean for planning the corrective action and retest, and what diagnostic method would you specify?
Read the worked answer
Convert the gap into airflow before scoping work: volume is roughly 96,000 ft2 times 10 ft of height, or 960,000 ft3, so 1 ACH50 equals about 16,000 cfm50. Closing 0.9 ACH50 means eliminating roughly 14,000 cfm50, which is far too much to attribute to a single parapet defect even where the leakage signal is strongest. The auditor should tell the client to expect several defects and budget for more than one repair cycle.
Specify diagnostic localization, not guesswork: a calibrated fan or guarded zone test combined with smoke-pencil or infrared survey of the top-floor and parapet zone under induced negative pressure, following the referenced air leakage detection method. Record the baseline conditions before the survey. Repairs should be sequenced, then a partial retest of the treated zone performed before a full whole-building retest. A single whole-building retest after one repair risks burning the test budget and returning another failure.
3. The owner asks whether the interior polyethylene should be removed while the wall is open. How should the auditor respond?
Read the worked answer
The interior polyethylene is not the cause of the observed condition, and removing it is not a substitute for repairing the air barrier. In this assembly, the exterior membrane is vapor permeable, so the wall dries primarily outward; a Class I interior retarder limits inward drying but is common and permitted in this climate zone when the air barrier is continuous.
The defensible position is sequence and evidence: repair and verify the air barrier first, keep the interior retarder intact unless an assembly-level moisture analysis shows it is the limiting layer, and never accept its removal in exchange for skipping leak repair. If the owner wants to delete it, that is a design change requiring the design professional's approval and a documented moisture analysis. The auditor should record the question and the recommendation in the report so the decision is traceable, and note that no double vapor retarder should be added as proposed.