Wind Damage Reseal and Hand Sealing

Purpose

After a wind event the shingles that came off are the smallest part of the problem. The ones still lying flat with a broken seal are the rest of it, and they are invisible from the ground. This procedure guarantees the roof is surveyed by hand rather than by eye, that a creased shingle is separated from a merely unsealed one, and that a slope-level decision gets made before anyone starts dabbing cement.

The failure this prevents is a crew hand-sealing forty tabs on a slope that had lost its seal generally, invoicing it as a repair, and having the same slope open again in the next storm with the shop's name on the last work order.

Scope

Covers post-wind survey and resealing on steep-slope asphalt shingle roofs, residential and small commercial: the hand-lift survey, classifying creased versus unsealed, the slope-level repair-or-replace threshold, and verification.

Does not cover the insurance-facing evidence pass, which the storm damage inspection and documentation SOP owns and which runs first. Does not cover the mechanics of lifting courses, fastener placement, cement pattern and the tug test, which the shingle repair and blend-in standard owns and which this procedure calls into. Ridge caps have their own SOP; so do emergency tarping and full slope replacement.

Roles and responsibilities

Two kinds of loose, and only one of them reseals

A tab that lifted in wind and lay back down is not damaged if the mat under the fold is intact. A tab that creased is: the fold has fractured the reinforcing mat, and the shingle will split on that line whatever you glue underneath it. Lift the tab and look at the underside at the fold - a crease reads as a pale straight line across the mat, usually with a matching line of missing granules on the face, and the shingle will not lie flat when you release it. Creased goes in the replacement count. Intact and unsealed goes in the reseal count.

Whether an intact shingle will re-bond on its own is a question about the sealant strip, not about the shingle. The factory strip bonds with heat, so a tab that broke free in January may lay back down and seal itself in July - unless dust, granules or oxidation have contaminated the strip, in which case it never will and no amount of sun helps. The strip is also where the wind rating lives: ASTM D3161 (fan-induced, classes A, D and F) and ASTM D7158 (uplift, classes D, G and H, the method the adopted IRC references for asphalt shingles) are both run on sealed specimens, so an unsealed tab does not hold a lower class, it sits outside the test. Assume contamination on any tab that has been open through a rain event, and hand seal it.

The threshold is a shop decision and it should be a number rather than a feeling. A workable default: once more than 20 percent of the shingles in a slope's survey sample lift with no resistance, or once unsealed shingles appear in more than one area away from the rakes, eaves and ridge, that slope is a replacement conversation rather than a hand-seal job. Tune it to your terms, but write it down.

Procedure

1. Set access and fall protection, and gate on the wind that is still blowing. Access runs under the ladder setup standard and the fall protection setup SOP. Acceptance: anchorage set, everyone connected, and a current wind reading or forecast the crew has agreed is workable. Wrong looks like a survey run the same afternoon the event is still passing. Stop rule: gusts that move loose material on the slope stop roof work, and the visit becomes a ground and ladder assessment. Hazard: a slope full of unsealed shingles slides underfoot in a way a sealed field does not, because each loose shingle can move on the one below - so plan foot placement on courses you have already tested, and treat the fall protection duty at 6 ft under 29 CFR 1926.501(b)(13) as covering the survey, not just the repair.

2. Document before you touch anything. The evidence pass runs first, because resealing a tab destroys the condition an adjuster needs to see. Acceptance: the documentation visit is complete and its photographs are in the file, or the customer has declined a claim in writing. Wrong looks like a crew cementing tabs on day one and an adjuster arriving on day four. Stop rule: no reseal work starts until one of those two is true. Hazard, commercial: the shop that destroys the evidence inherits the argument about whether the damage was ever there.

3. Survey by hand-lift on a defined sample, slope by slope. Lift by hand at the leading edge of every shingle in the eave course, the full rake column on both sides, and every third shingle on three evenly spaced mid-slope courses. Acceptance: a tested count and an unsealed count for each slope, recorded separately, with the sample pattern written down so it can be repeated. Wrong looks like a tech spot-checking wherever they are kneeling. Stop rule: a slope where the sample cannot be completed - too steep, too brittle, too much snow - is recorded as not surveyed rather than assumed sound. Hazard: hand-lifting on a cold roof cracks the mat you were testing, so the surface temperature window in the shingle repair and blend-in standard applies to the survey too, and a cracked shingle joins the replacement count.

4. Classify every unsealed shingle before any cement comes out of the pail. Lift and inspect the fold line on the underside, then release and watch whether it lies flat. Acceptance: every unsealed shingle assigned to one of three buckets - creased or torn, intact and contaminated, intact and clean - with a count in each. Wrong looks like cement under a creased shingle because it was quicker than replacing it. Stop rule: any doubt about a fold goes in the replacement bucket, because a resealed creased shingle fails at the fold and the shop owns it. Hazard: this work puts you face-down with both hands under a shingle, so lanyard length has to allow that position without shock-loading, and nothing is set down loose above you.

5. Apply the slope-level threshold before doing any work. Divide unsealed by tested for each slope, compare to the shop threshold, and say the number out loud to the customer. Acceptance: an unsealed percentage per slope, written, with the resulting decision recorded as repair or replacement conversation. Wrong looks like starting the reseal and mentioning the extent afterwards. Stop rule: a slope over the threshold does not get hand-sealed as a partial measure - it goes to the estimator as a slope, and if the customer wants interim protection they get it described as interim, in writing. Hazard, commercial: partial work on a failed slope reads to the customer as a completed repair, so the language on the invoice matters as much as the work.

6. Replace the creased, reseal the intact, and only inside the temperature window. Replace creased and torn shingles per the shingle repair and blend-in standard, then hand seal the intact ones and every shingle you unsealed to reach them, using the cement pattern and quantity that standard and the shingle wrapper specify. Acceptance: replacement count matches the step 4 creased bucket, dab count matches shingles sealed times the specified spots each, and surface temperature is at or above the minimum the shingle instructions state for hand sealing. Wrong looks like a continuous bead of cement instead of the specified spots. Stop rule: below the stated minimum temperature, sealing stops and reschedules rather than being done cold and re-done later. Hazard: uncured cement is both a rain exposure and a fire hazard - it is a solvent-carried product, so no heat source, torch or smoking near an open pail, and the label directions bind on cure time, so check the forecast against that time before the crew leaves.

7. Verify with a second sample, and write the negative for what you did not treat. After cure, re-lift a fresh sample on each treated slope, including shingles you sealed and shingles you did not, and tug-test every sealed tab. Acceptance: every sealed tab resists and returns flat, the second sample's unsealed count on treated slopes is zero, and any untreated slope is recorded with its own tested and unsealed counts. Wrong looks like a report that says the roof was resealed with no numbers behind it. Stop rule: any tab that lifts in the second sample is re-cemented and re-tested before the crew leaves the site. Hazard: this step restores the field's uplift resistance, which the survey itself disturbed on every tab it lifted, so the second sample is the only evidence the roof is not leaving in worse condition than it arrived.

The record this produces

A wind entry carrying, per slope: the sample pattern, the tested and unsealed counts, the unsealed percentage and the threshold it was compared to, the three classification buckets with counts, shingles replaced, shingles hand-sealed and total dabs, the surface temperature at sealing against the stated minimum, and the second-sample result.

The estimator reads the per-slope percentages, because that is the whole replacement conversation in one line. An adjuster reads the classification buckets, the only place creased and unsealed are separated with counts.

Worked pass: two slopes on a 6:12 hip, gusting event two days prior

Step 1: anchorage set, both techs connected, wind settled. Step 2: documentation visit complete, photographs in the file. Surface temperature reads 62 F, inside the working window.

The front slope is 30 ft wide with 18 ft of slope length: at 36 in wide and 5-5/8 in exposure, 10 shingles per course, and 216 divided by 5.625 is 38 courses, so 380 shingles. Step 3's pattern lands on 60 of them.

Step 5 fails on the front slope. Twenty-two of the 60 tested lift with no resistance, which is 36.7 percent, against the shop threshold of 20 percent, and the unsealed shingles are not confined to the rakes and eave. The stop rule fires: no hand sealing is done on that slope. It goes to the estimator as a slope replacement, and the customer is given interim protection described in writing as interim.

The rear slope samples 60 the same way and returns 6 unsealed, which is 10 percent, under the threshold, so the reseal proceeds. A full sweep of that slope finds 14 more, so 20 unsealed in total. Step 4 classifies them: 4 creased, 16 intact. Step 6 replaces the 4 creased per the shingle repair and blend-in standard, which requires lifting 5 shingles in the courses above them, and hand seals the 16 intact plus the 4 new plus those 5 lifted - 25 shingles at 4 spots each, so 100 dabs.

Step 7 re-lifts a fresh 30-shingle sample on the rear slope after cure, including 8 of the shingles sealed, and every one resists. All 25 sealed tabs tug-test flat. The front slope's numbers go in the file as an untreated slope with 22 of 60 unsealed.

The 36.7 percent is the number that made this visit honest. Every one of those 22 could have been dabbed and would have looked identical to the 16 on the rear slope, and the invoice would have said the roof was resealed.

References

• 29 CFR 1926.501(b)(13) and 1926.502(d), the fall protection duty and arrest criteria governing roof access for the survey as well as the repair
• ASTM D3161 and ASTM D7158, the wind-resistance test methods asphalt shingle classes are published under, both of which are run with the sealant bonded
• The shingle manufacturer's installation instructions for the product on the roof, which state the minimum temperature for hand sealing and the cement pattern and quantity; the roof cement label directions bind on cure time and handling
• See related: the storm damage inspection and documentation SOP, which runs before this one; the shingle repair and blend-in standard; and the ridge vent and ridge cap repair SOP