Envelope, air barrier and water testing

The building envelope, the walls, windows, doors and roof that separate the inside from the outside, causes more callbacks and claims than almost any other part of a building. Leaks damage finishes and structure, air leakage wastes energy and causes condensation, and failures often appear years after the contractor has left. Envelope defects are also hard to see: the critical layers, the water-resistive barrier, the air barrier and the flashings, are hidden under cladding within days of installation. This lesson covers how the Larkspur Commons team inspects and tests the envelope, building on the mock-up in lesson 4.

Many regions of the United States have learned hard lessons about envelope quality. Waves of moisture failures in wood-frame multifamily buildings, from the problems with face-sealed exterior insulation and finish systems in the 1990s to later claims involving windows and balconies, made construction-defect litigation a major cost in several states and pushed designers toward drained and vented rainscreen walls and owners toward envelope consultants' field reviews. Larkspur Commons is in the relatively dry Treasure Valley, but its owner adopted a rainscreen approach and envelope consultant reviews.

What the envelope must control

An envelope controls four flows: liquid water from rain and snow, air, water vapor and heat. Each has a control layer, and each control layer must be continuous across the whole envelope, including at the joints between materials and between trades. Most envelope failures are failures of continuity at joints: the window-to-wall interface, the wall-to-roof transition, balcony door thresholds, penetrations for vents and pipes, and the junctions where one trade's work meets another's.

On this project the sheet membrane on the sheathing does two jobs: it is both the water-resistive barrier and the air barrier. That makes its continuity doubly important. A gap at a window corner lets water into the wall and lets air carry moisture into the assembly in winter.

Inspecting the barriers before cover

The envelope ITP has inspections at each layer, with hold points before each layer is covered. The membrane is inspected after installation and before cladding girts and insulation go on; windows are inspected after installation and before the membrane strips that tie them in; the ties are inspected before cladding. The envelope consultant reviews a sample of each layer and has a hold point at the first work in each zone, as lesson 4 described.

Lapping direction is the classic error. Water runs downhill, so every lap must put the upper sheet over the lower one. At the head of a window, the head flashing goes over the window flange and under the membrane above. A strip installed in the wrong order looks neat and sealed, but it creates a reverse lap that funnels water behind the membrane. The envelope consultant's first two reviews on the wood frame found exactly this error at window heads, which is why lesson 2 described a witness point being added for the first window in each zone.

Adhesion and membrane checks

Self-adhered membranes depend on adhesion to the sheathing and to each other. Adhesion suffers if the surface is wet, dusty or too cold, or if the primer is wrong or skipped. Field checks include rolling all laps with a hand roller, probing lap edges with a blunt tool to find loose edges, and simple adhesion pull tests on sample areas, in which a strip is cut and pulled and the force or the failure mode is recorded. A membrane that peels cleanly off the substrate has not bonded; one that tears or leaves adhesive on the substrate has.

Temperature limits matter in a Boise winter. The membrane manufacturer's instructions set minimum surface and air temperatures, and the specification requires that they be followed. When temperatures fell in November, the air barrier crew switched to the low-temperature primer approved in lesson 5, and Mei added a daily check of the surface temperature with an infrared thermometer to the follow-up inspections.

Window water testing in place

The specification for Larkspur Commons requires field water testing of installed windows in addition to the mock-up test: a set of windows on each floor, chosen by the envelope consultant, is tested to ASTM E1105 at the specified pressure before interior finishes are installed. If a window fails, the cause is investigated and corrected, the window retested, and further windows of the same type tested. This sampling approach is common because testing every window is slow and expensive, while testing none leaves the owner relying on visual inspection of hidden work.

# Field window water testing by floor (illustrative)
# Specification: test 3 windows per floor; on any failure, test 3 more of that type on that floor
results = {
    "L3": [("W-A", "pass"), ("W-B", "pass"), ("D-1", "pass")],
    "L4": [("W-A", "pass"), ("W-A", "fail"), ("W-B", "pass"),
           ("W-A", "pass"), ("W-A", "pass"), ("W-A", "pass")],
    "L5": [("W-A", "pass"), ("W-B", "pass"), ("D-1", "fail"),
           ("D-1", "fail"), ("D-1", "pass"), ("D-1", "pass")],
}
total = sum(len(v) for v in results.values())
fails = sum(1 for v in results.values() for _, r in v if r == "fail")
for floor, tests in results.items():
    f = [t for t, r in tests if r == "fail"]
    print(f"{floor}: {len(tests)} tests, {len(f)} fail{'s' if len(f) != 1 else ''} {('(' + ', '.join(f) + ')') if f else ''}".rstrip())
print(f"overall: {total} tests, {fails} failures ({fails / total:.0%})")

L3: 3 tests, 0 fails
L4: 6 tests, 1 fail (W-A)
L5: 6 tests, 2 fails (D-1, D-1)
overall: 15 tests, 3 failures (20%)

Level 5's balcony door type D-1 failed twice, which was a pattern, not a one-off. The investigation found that the threshold sill pans on level 5 had been installed by a different crew, which had not attended the preinstallation meeting, and had omitted the end dams at the threshold ends. Every D-1 door on level 5 was opened up and corrected, and the crew went through a preparatory and initial phase as lesson 5 describes. The single failure on level 4 was a missing sealant plug at a frame joint, an isolated workmanship error.

Roofing inspection and testing

Larkspur Commons has a low-slope roof with a two-ply modified bitumen membrane over tapered insulation on the wood-frame roof deck, and a protected membrane on the podium deck over the parking garage, where the waterproofing sits under insulation, drainage board and pavers. Roof inspection during installation checks the substrate, the insulation fastening or adhesion, the membrane laps and seams, flashings at curbs, parapets and drains, and the details at penetrations. Seams are probed for voids and the manufacturer's representative usually inspects the roof for the warranty.

The podium deck was flood tested in sections before the drainage board and pavers went on, because a leak found after the pavers are placed means removing them, and a leak over the parking garage drips onto cars. The roof over the residential floors was tested by electronic leak detection, which found two small breaches: a puncture from a dropped tool near a mechanical curb and an unsealed seam end at a drain. Both were repaired the same day. Lesson 1 mentioned a podium waterproofing leak that occurred during the first heavy rain; it was at a wall-to-deck transition that had not been included in the flood test area, and the ITP was changed so that every flood test area includes its perimeter transitions.

Air leakage testing

Air tightness can be measured. A whole-building or whole-zone air leakage test uses large fans to pressurize or depressurize the building and measures the air flow needed to hold a set pressure difference across the envelope. The result is normalized by the area of the envelope and reported, for example, as the air flow per square meter of envelope at 75 Pa. Test standards such as ASTM E779 and ASTM E3158, which addresses large and multizone buildings, describe the procedures, and Canadian practice also uses the CGSB 149 series of standards.

Airtightness testing is required in some US jurisdictions through the energy code's compliance options, and many owners and funders of multifamily housing require it. For Larkspur Commons, the owner's specification requires a whole-building test with a target of 0.30 cfm per square foot of envelope at 75 Pa, along with compartmentalization tests on a sample of dwelling units to limit air, odor and smoke transfer between homes. The target is a project requirement chosen by the owner, not a code minimum.

A good test is done at several pressures, not just one. The relationship between flow and pressure through the many small cracks in an envelope follows a power law, so the testing agency measures flow at a series of pressures, fits the curve and reports the flow at the reference pressure. The script below shows the calculation for a preliminary test of level 3 west, done before drywall so that leaks could still be found and sealed.

# Multipoint air leakage test, level 3 west zone before drywall (illustrative)
import math
area = 12700.0  # ft2 of exterior envelope enclosing the zone, plus the boundary to adjacent zones
data = [(25, 2080), (35, 2610), (45, 3070), (55, 3520), (65, 3920)]   # (pressure Pa, flow cfm)
xs = [math.log(p) for p, _ in data]; ys = [math.log(q) for _, q in data]
n = len(data); mx = sum(xs) / n; my = sum(ys) / n
slope = sum((x - mx) * (y - my) for x, y in zip(xs, ys)) / sum((x - mx) ** 2 for x in xs)
C = math.exp(my - slope * mx)
q75 = C * 75 ** slope
print(f"flow exponent n = {slope:.2f}, coefficient C = {C:.1f}")
print(f"flow at 75 Pa = {q75:.0f} cfm, normalized = {q75 / area:.3f} cfm per ft2 (target 0.300)")

flow exponent n = 0.66, coefficient C = 246.3
flow at 75 Pa = 4316 cfm, normalized = 0.340 cfm per ft2 (target 0.300)

The zone came in above the target. That was not surprising at a preliminary stage, and it was the reason for testing early. While the fans held the zone under pressure, the testing agency and Mei walked the zone with a smoke pencil and an infrared camera, which showed cold air streaming in at three kinds of location: the joints between the floor framing and the wall membrane at the rim, a set of unsealed electrical boxes in exterior walls, and the gaps around exhaust duct penetrations. All three were sealed and the zone was retested at 0.28 cfm per ft², below the target. The lessons were built into the ITP for the remaining zones, and the final whole-building test passed with margin.

The flow exponent says something too. Values near 0.5 suggest a few large openings, where air behaves like flow through an orifice; values near 1 suggest many tiny cracks. A typical building envelope falls between about 0.6 and 0.7. A result well outside that range is a reason to check the test setup.

Finding leaks: diagnostic tools

When a test fails, the problem is finding where. The common tools are simple and effective in experienced hands.

• Smoke pencils or theatrical smoke show air movement at a crack while the building is pressurized.
• Infrared thermography shows temperature differences caused by air leakage or wet materials, best when there is a large indoor-outdoor temperature difference.
• Calibrated spray racks and hoses find water leaks by testing small areas in sequence, starting low and working up, so a leak is not masked by water from above.
• Moisture meters measure the moisture content of wood framing and sheathing, used before closing walls and in investigating leaks.

On wood-frame buildings, moisture content of the framing before close-in is itself a quality check. Wet framing closed into walls can shrink, cause cracks and nail pops, and grow mold. The Larkspur Commons specification sets a maximum moisture content for framing before insulation and drywall, and Omar's moisture readings are a review point in the insulation ITP (CON-335 lesson 3).

Coordination between trades

Envelope work involves more trades than almost any other part of the building: framers, membrane installers, window installers, cladding installers, roofers, sheet metal workers, electricians and plumbers for penetrations, and the mechanical trade for vents. Each can damage another's work. Cladding installers puncture the membrane with fasteners in the wrong places; electricians cut holes for exterior lights and do not seal them; roofers and sheet metal workers argue about who flashes the parapet. The preinstallation meeting in lesson 4 settled the main interfaces, but day-to-day protection needs a clear rule. On Larkspur Commons, any trade that penetrates the membrane after it has been inspected must use an approved seal and record the penetration on the zone's envelope log, and the log is checked before cladding.