Curtain Wall Mockup Air Leakage: Corrections Before You Condemn the Assembly
A unitized curtain wall mockup 4.8 m wide by 7.2 m tall - two floors with a horizontal splice - is erected in an independent laboratory for air leakage testing per ASTM E283 at 75 Pa. The specification limits fixed glazing to 0.30 L/(s·m²), with operable units limited separately. The first run reads 31 L/s total, which is 0.90 L/(s·m²) over the 34.6 m² specimen. Laboratory conditions are 32 °C air and 84.6 kPa barometric pressure at an elevation of 1,450 m. During the post-run walkdown the lab technician finds two 50 mm conduit sleeves in the chamber wall standing open to the laboratory, and the mockup installer admits he taped two vertical joints 'because they rattled' before the run. The lab's nozzle-array flow measurement carries ±5% uncertainty. The project manager asks whether the assembly failed and whether the mockup has to be rebuilt.
1. Before declaring failure, what corrections and checks apply, and what is the corrected result?
Read the worked answer
Three things must be dealt with, and one of them turns out to explain almost the whole reading.
Density correction. At 32 °C and 84.6 kPa, ρ = P/(R·T) = 84,600/(287 × 305.2) ≈ 0.966 kg/m³ against a standard 1.20. At fixed Δp, volumetric leakage scales as 1/√ρ, so q_std = q_meas × √(ρ_test/ρ_std) = 0.90 × √(0.966/1.20) = 0.90 × 0.897 ≈ 0.807 L/(s·m²). Real, but nowhere near enough to save the specimen on its own.
The open sleeves. Two 50 mm penetrations = 2 × 0.00196 = 0.0039 m². Treating them as sharp-edged orifices with Cd ≈ 0.61: q = Cd·A·√(2Δp/ρ) = 0.61 × 0.0039 × √(150/0.966) = 0.61 × 0.0039 × 12.46 ≈ 0.030 m³/s, or about 30 L/s. The measured total was 31 L/s. The chamber penetrations, not the curtain wall, account for essentially the entire result.
The tape. Taping joints reduces leakage, so the run is simultaneously pessimistic (sleeves) and optimistic (tape). Either way it is not a measure of the as-built assembly.
Compare against the ±5% uncertainty: 31 L/s ± 1.6 L/s. The result cannot be adjudicated on a test with uncontrolled chamber penetrations.
2. What should be reported, and what is the re-test protocol?
Read the worked answer
Report: the run as invalid for compliance purposes, with the raw flow, the corrected flow, the test conditions, the correction factor applied, the ±5% uncertainty, the chamber penetrations, and the installer's tape. An invalid run is documented, not deleted - the record shows what happened and why the retest is credible.
Re-test protocol:
- Remove all tape and any other temporary seals, and walk the specimen with the installer to confirm nothing else has been added.
- Seal the conduit sleeves and verify the seal.
- Run a chamber tare or blank-off test - seal the specimen face and confirm the chamber itself reads essentially zero at 75 Pa. This is the check that should have preceded the first run and is the single most useful habit in mockup testing.
- Confirm the nozzle-array calibration and record temperature and barometric pressure at the time of test.
- Re-run at 75 Pa, apply the density correction, and document it. Where the specification or standard calls for a multi-point series, run the series rather than a single point so the flow exponent and correlation can be reported.
Only after that is there a basis for saying the assembly passes or fails.
3. If the corrected result still exceeds the limit, how do you decide whether the mockup or the test setup is at fault, and where do you look?
Read the worked answer
Allocate before you rebuild. A single whole-specimen number tells you the total, never the location.
Bisect the specimen. Temporarily baffle the mockup internally - split the fixed lights from the operable sashes, and the lower floor from the upper - and measure each sub-area at 75 Pa. If one zone holds most of the leakage, the problem is that zone's details, not the system.
Test components separately where the specification allows: a single fixed light, a single operable sash, a single horizontal splice.
Check the interfaces first. In mockup testing the most common false failures are not the product at all: they are the chamber-to-specimen perimeter, the mockup-to-floor junction, and instrument and power penetrations. Confirm the perimeter seal and repeat the tare check.
Look at what the installer touched. He taped two vertical joints because they rattled. Rattling means those joints were moving and open under pressure. That is where the leakage is likely to be, and it is also a design or fabrication finding worth feeding back to the supplier, not just a field fix.
Then remediate and re-test the same mockup. Rebuilding before you know which component leaks is expensive guesswork.