Sampled Airtightness Test in a 48-Unit Apartment Block with a Gusting Wind Baseline
A testing agency is engaged to verify envelope airtightness for a new 48-unit, six-storey wood-frame apartment building. The specification requires 2.50 ACH50 or less using ASTM E779, but the owner's consultant proposes sampling: 12 units tested individually at 50 Pa, with the units above, below and on both sides of each test unit treated as guard units. On test day the site is windy - 4.5 m/s sustained, 7 m/s gusts at 10 m - and the indoor-to-outdoor temperature difference is 24 °C. With the fans off, the technician logs baseline pressures of -2.1, +7.9, -9.4 and +6.2 Pa across a 60-second sample, then confirms that the pre-test and post-test baseline averages differ by 2.8 Pa. The fan set reaches exactly 50 Pa in unit 101 and holds 145 L/s; the unit's conditioned volume is 212 m³. The technician computes 2.46 ACH50, writes 'PASS' in the field log, and moves on to the next unit. A second technician questions the result at lunch.
1. Is unit 101's 2.46 ACH50 a reportable result? Work through the baseline requirement and show the arithmetic.
Read the worked answer
The arithmetic is right: 145 L/s × 3.6 = 522 m³/h, and 522 ÷ 212 m³ = 2.46 ACH. Correct arithmetic does not make the test valid.
ASTM E779 (confirm the edition your specification invokes) requires the baseline pressure signal - the wind and stack effect measured with the fans off, before and after the test - to be small relative to the test pressure, conventionally no more than 10% of it. At 50 Pa that allowance is 5 Pa.
The four logged readings average 6.4 Pa in absolute value: (2.1 + 7.9 + 9.4 + 6.2) ÷ 4 = 6.4 Pa. That is 12.8% of 50 Pa. Both the mean-of-absolutes test and the pre/post-pair test have to be satisfied; a 2.8 Pa drift between the two baseline sets also says the wind was not steady during the test. With 7 m/s gusts the fan is partly fighting the wind, so even a spot reading at exactly 50 Pa can hide real variation across the specimen.
The correct entry is 'invalid - baseline out of tolerance,' not 'PASS.' Because 2.46 sits only 1.6% under a 2.50 limit, any bias in the measurement is larger than the margin.
2. What are the legitimate options for finishing the test, and which would you recommend? Include what must appear in the report.
Read the worked answer
Options:
- Wait and re-test. Return when the wind is calmer and the indoor-outdoor temperature difference is smaller. This is the cleanest answer but may cost days on a 12-unit sample.
- Raise the test pressure. The same 6.4 Pa baseline is only 8.5% of 75 Pa, inside the 10% allowance. Run a multi-point test up to 75 Pa and evaluate the curve at 50 Pa. This is legitimate if the fan capacity allows and the results are documented as an extrapolation.
- Average longer. Longer sampling reduces noise, not bias. It cannot convert a 12.8% baseline into a valid 50 Pa point and should not be presented as a fix.
Recommended: a multi-point test (the standard's series of points from roughly 10 Pa up to the highest attainable pressure), fitting q = C·Δp^n, reporting the flow exponent n, the correlation coefficient, and the value evaluated at 50 Pa. The report must show: all baseline readings before and after, the test pressure and flow for each point, the indoor/outdoor temperature and barometric pressure used for the density correction, the definition of volume used for ACH (conditioned space), and an explicit note that the 50 Pa value was obtained from the curve rather than measured directly.
3. For the next unit, the corridor doors were locked and none of the guard units could be connected to the test fan. Explain what that does to the measured leakage.
Read the worked answer
Guard units exist to drive the pressure across shared assemblies to approximately zero. Party walls, the floor above and the floor below are interior boundaries; they are not part of the enclosure. If a guard unit is left at ambient while the test unit sits at -50 Pa, every one of those boundaries sees the full 50 Pa and any leakage through it is supplied by the measuring fan. The fan has no way to distinguish that air from air crossing the exterior skin, so it is all booked as envelope leakage.
In platform-framed wood construction the party-wall and floor-to-floor paths can be a large fraction of total leakage - far more than the 0.04 ACH of margin unit 101 had. The unit would appear to fail a 2.50 limit for reasons that have nothing to do with the enclosure, and the fan would have to work noticeably harder than the 145 L/s logged.
The remedy is procedural: open all guard units and hold them at the test-unit pressure (the project protocol usually sets a tolerance, commonly a few pascals or a percentage of test pressure), then verify the guard-to-test-unit differential with a second gauge before starting. If the guards cannot be set up, the test is not run - do not report the number with a footnote.