Field Air Infiltration Test Technique for New Window Installs

Why this matters

A window can be foamed, flashed, and trimmed exactly to the standard and still leak air, because a missed corner in a bead of sealant or a compressed spot in a weatherstrip does not show up on a visual inspection. The energy code compliance and NFRC labeling references cover what number the unit is rated for and what threshold the jurisdiction requires; neither one tells you whether the specific opening in front of you, on this job, actually achieved it. This guide is that field test: how to create a real, repeatable pressure differential across a completed opening and trace exactly where air is still moving, before the job closes rather than after a customer calls back cold.

Before you depressurize anything: the hazard this test creates

Any test that pulls the house negative relative to outside can backdraft a natural-draft combustion appliance, gas water heaters and older atmospherically-vented furnaces are the common ones, pulling combustion gas including carbon monoxide back into the living space instead of up the flue. Before running this test, identify every combustion appliance in the house. Where a furnace or water heater is sealed-combustion or direct-vent, it draws and exhausts through its own dedicated piping and is not at meaningful risk from house depressurization; where it is natural-draft, either shut it down for the duration of the test or place a calibrated carbon monoxide monitor near it and watch it continuously while depressurized. If the monitor's reading climbs at any point during the test, stop immediately, open doors to depressurize the house back to ambient, ventilate the appliance space, and do not restart the appliance or the test until the space has cleared and the cause is found. The same depressurization also pulls air from an attached garage into the house through any shared wall penetration; if a vehicle has run in the garage recently or fuel and solvent containers are stored there, treat that pathway the same way and monitor it rather than assuming the garage is sealed just because it has a door.

What this test is, and what it is not

The reactive version of this check, run in response to a customer's draft complaint, uses an incense stick or tissue on a cold, windy day because natural stack effect and wind are the only pressure differential available after the fact. That works for a one-time diagnosis but it is not repeatable; a calm day gives a false pass. This guide's technique creates its own pressure differential mechanically, with a blower door or an equivalent calibrated exhaust fan sealed into a doorway, so the test produces the same result on a still July afternoon that it would on a gusty January morning. That repeatability is what makes it a commissioning check the shop can run on every job rather than a diagnostic that only works when the weather cooperates.

A specification-driven commercial job may call for a full quantitative test under ASTM E783, Standard Test Method for Field Measurement of Air Leakage Through Installed Exterior Windows and Doors: a sealed chamber built over the single unit, a calibrated fan holding the same static pressure differential the unit's lab rating was measured at, commonly 75 Pa (1.57 psf), and a flow measurement converted to cubic feet per minute per square foot of the tested area, compared against the specified maximum. That test needs equipment most window-replacement shops do not own and is normally run by a qualified field-testing agency; know when a spec calls for it and route the job there rather than improvising a substitute and calling it equivalent. What the rest of this guide covers is the qualitative whole-house or whole-room technique a shop's own blower door supports, which locates leak paths precisely even though it does not produce a certified cfm-per-square-foot number.

Setting up the differential

Seal the blower door frame into an exterior doorway on the same level as the tested windows, close all other exterior doors and windows, and set interior doors to their normal in-use position rather than propping every room open, since an open interior door changes which rooms actually see the test pressure. Run the fan to depressurize the house and aim for the 75 Pa standard test condition where the equipment and the house's overall leakage allow it. On a large or already-leaky house the fan may not hold 75 Pa stably; where that happens, run the test at the highest stable differential the fan actually achieves and record that number rather than the target, because a test run at an unrecorded and unstated pressure cannot be compared to a retest six months later.

Tracing the leak with smoke

Use a non-toxic smoke pencil or squeeze-bulb puffer, not a heat-based smoke source, since an open flame near a depressurized house is its own ignition concern and is unnecessary here. With the house holding its stable differential, work the smoke source slowly along the interior perimeter of each tested window: the head, both jambs, the sill, and the interior casing-to-wall joint, in that order, holding the tip within about an inch of the joint at each point and watching for smoke pulled into the gap rather than drifting past it in still room air. A true leak point pulls the smoke stream consistently, not once on a lucky pass; work each point slowly enough to confirm it before moving on. Document a pass or a fail at each of the five points per unit; a single missed corner on an otherwise sealed perimeter is exactly the kind of leak this test exists to find and a casual once-around will not catch.

Reading a failure back to its cause

A leak at the sill or a jamb traces most often to a gap the trim finish standard's perimeter foam fill should have closed; go back to that step rather than reaching for exterior sealant first, since sealing the outside over an unfilled interior gap treats the symptom and leaves the air path intact behind it. A leak at the interior casing-to-wall joint traces to a skipped corner in that same standard's continuous interior sealant bead, the specific failure its own stop rule exists to catch; if the trim has already been painted, that skip may be sitting under an unbroken-looking paint film, bridged rather than actually filled. Re-open, re-seal, and retest that single point rather than re-sealing the whole perimeter; indiscriminate re-caulking around a window risks sealing a weep opening the drainage plane depends on.

Worked example: three bedroom windows, post-install commissioning check

A 1994 ranch had three bedroom windows replaced as part of a job carrying a utility rebate that required documented as-installed air-sealing verification. The house runs a sealed-combustion direct-vent furnace and an electric water heater, so backdraft risk from those two appliances was low; a carbon monoxide monitor was still placed in the mechanical closet and watched through the test as standing practice, and it never moved off zero. The attached garage shares a wall with the mechanical closet, so that pathway was also watched; no odor or monitor response there either.

The blower door held a stable 50 Pa differential; the equipment could not sustain the full 75 Pa target on this house's overall leakage, so 50 Pa was recorded as the actual test condition rather than the unmet target. Each of the three windows was scanned at five points: head, left jamb, right jamb, sill, interior casing-to-wall joint, for 15 scans total.

The first window, in the room referred to on the log as Room A, passed clean at all five points, no deflection anywhere. The second window, Room B, deflected at one point, the sill, while the other four passed. The third window, Room C, deflected at one point, the interior casing-to-wall joint, while its other four passed. That is two failing points out of 15 scanned, both isolated to a single point on their respective units rather than spread across multiple points, which is the pattern of a specific missed spot rather than a systemic install problem.

Room B's sill was traced to a gap at the corner where the sill pan's back dam meets the jamb, a spot the perimeter foam fill had bridged over rather than filled. That corner was re-foamed per the trim finish standard's own step, allowed to cure, and rescanned: no deflection. Room C's interior joint was traced to a corner the sealant bead had skipped, already covered by the finish coat of paint. That corner was reopened, resealed, and the paint touched up, then rescanned: no deflection. All 15 scan points passed on the completed retest, with the two initial failures and their fixes recorded in the same log rather than only the clean final result, since the failing points are what a rebate auditor or a future warranty call will actually want to see was caught.

Verify and close the record

The completed test log is the verification; there is no separate walkthrough beyond confirming every scanned point on the log shows a pass, including any point that failed and was corrected, with its retest result entered rather than left blank. Record the actual differential achieved, not the target, the CO monitor's reading confirmed at zero or logged if it moved, and the pass or fail result at each named point per unit. That log is what answers a rebate program's documentation requirement and, later, a warranty desk's question about whether a specific corner was ever actually checked.

References

• ASTM E783, Standard Test Method for Field Measurement of Air Leakage Through Installed Exterior Windows and Doors, for the quantitative chamber-and-fan technique on specification-driven work.
• AAMA 503, Voluntary Specification for Field Testing of Newly Installed Storefronts, Curtain Walls, and Sloped Glazing Systems, for the light-commercial equivalent of a field air and water test.
• Building Performance Institute blower-door and combustion-safety testing standards, for depressurization technique and backdraft precautions.
• See related: the interior and exterior trim finish SOP, which owns the perimeter foam fill and interior sealant bead this test verifies; the troubleshooting decision tree for a customer-reported draft after install, which owns the reactive single-visit diagnostic this guide's proactive commissioning check is distinct from; the NFRC ratings interpretation and energy code compliance references, which own the rated air-leakage figure and the code threshold this field result is checked against.