The Tarp That Sailed Off the Open Roof
Why this matters
An opening left in a roof overnight is a promise to the customer that the house will be dry in the morning. When the cover fails, the shop owns every ceiling, every wet batt of insulation and every hour of drying, and it learns about it from a phone call before the office opens. This case follows one of those calls from the first minute: what was done to stop the damage before anyone thought about cause, what the roof showed once it was safe to look, and why the tie-and-brick method that had worked on dozens of calm nights was never a cover at all.
The job and the night
A three-person HVAC crew was replacing a rooftop package unit on a one-story house with a low-pitch gable roof. The old curb was removed and the opening enlarged for an adapter curb, which left a hole through the roof deck into the attic, a little over 3 by 4 feet, about 4 feet below the ridge on the south slope. The new curb would not arrive until morning.
At 4 p.m. the crew covered it with a 12 by 16 foot light poly tarp. Grommets on the north and south edges were tied with cord to the old curb framing that remained at the opening. The east and west edges were held down with eight bricks from the customer's yard, four per side.
The evening forecast the lead had looked at said winds 8 to 12 mph overnight. It also said, in the hourly gust column, gusts to 35 mph after 2 a.m. as a front came through, and rain from about 1 a.m. He had read the sustained figure.
6:05 a.m.: stop the water, then think
The homeowner called the on-call line. Water was dripping from the hallway ceiling and a bedroom ceiling, and there was a soft bulge in the drywall near the hall light. The on-call tech's first questions were not about the roof.
What the on-call tech told the customer on the phone, in this order:
- Keep everyone out of the hall and that bedroom, and do not stand under the bulge. Wet drywall carrying water can let go all at once.
- Do not touch the hall light switch or the fixture. Water near a fixture is a shock risk.
- If the electrical panel is in a dry spot and you can reach it standing on a dry floor, switch off the breakers labeled for the hall and that bedroom. If the panel area is wet, or the labels are unclear, leave it and wait for us.
- Put buckets or pots under the drips and move anything you care about out of those two rooms.
On arrival at 6:50, with gusts still near 30 mph and steady rain:
- Power first. The tech confirmed at the panel that the two circuits were off and the customer had picked the right ones, then checked the hall fixture and the bedroom outlets with a non-contact voltage tester, testing the tester on a live outlet in the dry kitchen before and after. Nothing wet was touched until it read dead. The two breakers stayed off and taped until the fixture box and the bedroom outlets had dried and an electrician had looked them over; only then were they switched back on one at a time, with the hall light and each outlet confirmed working and neither breaker tripping.
- The bulge. Standing beside it rather than under it, wearing eye protection, with a bucket held under the low point, he made a small hole at the bottom of the bulge with a screwdriver and let it drain. That turns an uncontrolled collapse into a controlled drip. The bulge was nowhere near the fixture box, which is why it was done at that point and not at the light.
- The attic, not the roof. The tech did not go up on the roof in rain with 30 mph gusts and no way to secure himself or a tarp, and the shop's roof-walk hold would not have allowed it. He went into the attic instead with a headlamp, stepping only on the joists, and laid a scrap tarp and two buckets on the insulation below the opening to catch the water before it reached the ceiling. That cut the inside drip to almost nothing within twenty minutes.
- The office was told by 7:15 so the owner could call the shop's insurance agent that morning and get a drying contractor on the schedule. Wet insulation and drywall that sits wet grows mold, and the drying clock starts when the water stopped, not when the claim gets filed.
By 9:30 the gusts on the hourly had dropped to the high teens and the rain had stopped. Two techs went up with fall protection, rigged to the shop's standard for that roof, and only then did anybody look at why it happened.
What the roof showed
The tarp was in the backyard, wrapped over the fence. On the roof:
- The north-edge grommets had torn out of the tarp. The cord ties were still knotted to the curb framing, with a ragged ring of tarp in each loop.
- On the south edge, the ties held and the tarp ripped beside them, in a line from grommet to grommet.
- The bricks were gone from the west side. Two were on the roof, one in the gutter, and one in the side yard about 15 feet from the house, which is a brick that was thrown.
- The east-side bricks were still on the roof, slid downslope to the eave.
The sequence reads cleanly from that. The front's gusts came from the west. The west edge, held by four bricks, lifted first, and once air got under the tarp the whole sheet became a sail pulling against eight grommets. The grommets on a light poly tarp are a reinforced hole in a thin sheet, and each one carries its share of the entire sail's pull at one point. They failed one after another, and rain had a 3 by 4 foot hole into the attic for four to five hours.
The force involved is not subtle. Wind's velocity pressure, in the form ASCE 7 uses for structural wind loads, is about 0.00256 times the speed squared, in pounds per square foot with speed in mph, so a 35 mph gust carries about 3.1 pounds per square foot. That is per square foot of exposed sheet. If even a tenth of the 192 square foot tarp lifts, about 19 square feet, the pull on it is around 60 pounds, and it grows as more of the sheet fills. Eight bricks at roughly 4 to 5 pounds each are 32 to 40 pounds in total, and they are not even resisting together: each brick holds only the patch of tarp beneath it, on a wet sheet it can slide across. That figure is rough, since a flapping sheet does not load evenly, but it does not have to be precise to show the method was never in the same range as the load.
The crew's method had worked before because it had never met a gust. That is the point of the hourly's gust column, and it is the subject of Sustained Wind Versus Gusts and Which One a Limit Means.
Why battens and a ridge wrap would have held
The re-tarp at 9:30, which went on to survive a second windy night before the curb was set, was built differently in three ways.
A rigid lid under the tarp. A sheet of plywood was screwed to the remaining curb framing over the opening first. The tarp is then a weather skin over a structural cover rather than the only thing over a hole, and if it does tear, the rain meets wood instead of attic.
Battens instead of grommets. Wood furring strips were screwed through the tarp into the roof deck along every edge, with the tarp edge rolled once around the strip so the screws pass through doubled material. A batten clamps the whole edge continuously, so the wind's pull is spread along its length instead of being concentrated at a grommet every few feet. Screw holes through shingles were placed inside the area the new curb's flashing would cover, with any outside it noted for sealing when the roofing work was finished.
The upslope edge over the ridge. The tarp was run up from below the opening, over the ridge, and about 4 feet down the north slope, where it was battened. That does two things. Water running down the roof from the ridge flows over the tarp, the way an upper shingle laps a lower one, instead of under its top edge. And there is no exposed top edge for a wind coming over the ridge to get under, which is where a tarp that is battened on all four sides on one slope still tends to start lifting.
Side section, south slope, ridge at top (not to scale)
north slope RIDGE south slope
/\
batten, 4 ft down / \
on far side -> = \ tarp over ridge
/ \
/ \= batten, top of lid
/ [==] \ plywood lid on curb
/ opening \= batten, below lid
/ \
/ \= batten, low edge
EAVE EAVE
A batten also ran across the middle of the tarp below the lid to stop the sheet between edges from ballooning, which is how a well-battened tarp still tears: it flogs in place until the material fatigues.
The overnight protection standard that followed
The shop wrote this as the rule for any opening left in a roof or wall overnight, and moved the full procedure into The Overnight Open-Work Protection SOP:
- Nothing loose on a roof overnight. No bricks, blocks or buckets as ballast; they do not hold, and when they leave they are projectiles.
- A rigid lid over any opening larger than a hand, fastened to framing, before the tarp goes on.
- Battens on every edge, screwed at about 12 to 16 inch spacing, with the tarp rolled around the batten.
- The top edge goes over the ridge where the opening is within reach of it, battened on the far side; where it is not, the top edge is tucked under the course of roofing above the opening.
- The lead reads the gust column, not the sustained column, for the whole night, and a forecast gust above the shop's open-roof threshold means a second person checks the tarp before the crew leaves.
- A photo of the finished cover goes on the job record before the truck leaves.
The method itself, with materials and fastener choices, is in How to Tarp an Open Roof So It Stays Put Overnight. What this case adds is the order of the first hour, which no tarp method covers: power off and proved dead, the bulge drained from the side, water caught in the attic, the insurance and drying call made early, and the roof left alone until the wind allowed it.
References
- Wind velocity pressure relationship as used in ASCE 7 for structural wind loads, cited here only for the speed-squared relationship, not as a design method for temporary covers.
- See related: Sustained Wind Versus Gusts and Which One a Limit Means
- See related: How to Tarp an Open Roof So It Stays Put Overnight
- See related: The Overnight Open-Work Protection SOP
- See related: How to Secure a Job Site Days Before a Forecast Storm