Fastening a Roof for Wind Uplift

The Only Dimension the Whole Layout Hangs On

Zone width has one symbol, a, from the same rule in every edition that uses it: the smaller of a tenth of the least horizontal building dimension and four tenths of the mean roof height, with a floor of four percent of the least dimension and an absolute floor of three feet. Here a tenth of 46 m is 4.6 m and four tenths of nine metres is 3.6 m, so height governed and a is 3.6 m — just under twelve feet. Put the deck at six metres instead and a drops to 2.4 m, moving the whole chalk layout.

The word least is the one that gets misread. It is the SHORTER plan dimension that sets a, so a long narrow building carries narrow zones running its entire length rather than the generous ones its footprint suggests. On an L-shaped or stepped plan take the least dimension of the rectangle enclosing the roof under consideration, and treat a roof at a different level as a separate roof with its own height and its own a — the step between two levels is an edge for the upper one and an obstruction for the lower.

Pressure comes next and the arithmetic is short. Velocity pressure at mean roof height is 0.00256 × Kz × Kzt × Kd × Ke × V² in psf, with V the mapped basic wind speed in mph at the building's risk category. For components and cladding the directionality factor Kd is 0.85, and an enclosed building carries an internal pressure coefficient of plus or minus 0.18 added to whatever external coefficient you read. On this roof — 115 mph, Exposure C, nine metres to the deck — Kz works out near 0.98 and the velocity pressure near 28 psf, about 1.35 kPa, the same everywhere on the roof.

What is not the same everywhere is GCp, and this site does not supply it. It is read from the components-and-cladding figure in the code edition your jurisdiction has actually adopted, against roof form, slope, zone, and the effective wind area of the thing being held down — which for a single fastener and its plate is a small area and therefore a large coefficient. Editions differ in how finely the low-slope figure is cut: older ones give three zones, newer ones subdivide the edge and the corner further and add an interior zone for very large roofs. Read the figure you are held to, not the one you memorised. The corner coefficient on any of them is a multiple of the field value, not a few percent above it.

One more conversion decides whether the schedule is right or off by a factor. Wind pressures from ASCE 7-10 onward are strength-level, and the 0.6 in the allowable-stress load combinations is what brings them down to the level a service-load capacity is compared against; mixing the two silently halves or doubles a fastener count. Kzt is the other quiet one — taking it as 1.0 is correct on flat ground and wrong on a hill, a ridge or an escarpment, where speed-up is a separate calculation.

1. Establish mean roof height and the least plan dimension from the drawings, then confirm both on site — a re-roof that has already gained 150 mm of build-up is taller than the record set.
2. Compute a, and double it; those are the two offsets you will chalk.
3. Snap the a line continuously around the whole perimeter, inside the base flashing, before any board is placed.
4. Square the corner boxes off both adjacent edges rather than mitring them, and mark each with a paint dot the boards will not hide.
5. Mark every curb, hatch, skylight and equipment stand in the field — each is a local edge in its own right.
6. Photograph the chalked roof from each corner — once insulation covers the lines, those frames are the only record.

Rows, Not a Grid

A mechanically attached single-ply roof is not fastened on a grid, and picturing it as one leads to the wrong takeoff every time. The fasteners live in the seam. A sheet is rolled out, its edge is fastened in a straight row through the lap, and the next sheet is laid over that row and welded or taped to it, which means the plates are buried under the weld and the exposed field of every sheet carries no fasteners at all. Row spacing and sheet width are therefore the same number, and choosing one has chosen the other.

That is what makes zone enhancement a layout decision rather than a fastening decision. To tighten the pattern in the perimeter you do not add screws to a 3 m sheet; you cut the sheet down — to two metres, or to half width, or to whatever the manufacturer's wind table names for the pressure you computed — and you get more rows because you have more seams. Fasteners within the row are spaced at the table's figure as well, and the two together give the density the assembly was listed at. Set the corner sheets out first, while the roof is still empty; they are the fiddliest on the job and the ones a tired crew fudges.

Cutting sheets changes the material order in two directions at once. Narrower sheets mean more linear metres of seam for the same area — more weld time or more tape, and offcuts unless the sheet is a stock half-width — and they mean the roll count for the perimeter strip is not the roll count for the same area of field. Work the layout against the real roof geometry, the curbs and drains and equipment stands, rather than against a bare area figure.

Two habits sit behind most mechanically attached failures that are not design failures. The first is the screw set at the wrong depth: overdriven, the plate dishes and cuts the sheet at its rim; underdriven, it stands proud and telegraphs, and every foot that lands on it works the fastener loose. Set the gun's depth on a test board each morning and again when the deck warms. The second is a row that drifts off the flute — a line that wanders half a flute over twenty metres has a run of fasteners hanging in a void with a satisfying driven torque and nothing holding them.

The Boards Under the Sheet Keep Their Own Schedule

Insulation and cover board have a fastening density of their own, and it is not the membrane's. It is expressed as fasteners per board, and it steps up in the perimeter and corner exactly as the membrane's does — a fact that gets lost because the boards go down first, before anyone is thinking about wind, and because the person placing them is not the person who read the wind table. Give the board crew the same chalk lines and the three patterns drawn on a board offcut nailed to the parapet, in pictures rather than numbers.

Board layout interacts with the zone lines in a way the membrane's does not. A 1.2 by 2.4 m board straddling the a line is in two zones, and there is no such thing as fastening half a board to one pattern: take the higher pattern for any board crossing a line, or set the layout out so joints fall on the line. Where the assembly is two layers, keep joints in the upper layer off the joints in the lower — a coincident joint through both is a hinge running straight down to the deck, and it is where a corner starts to lift.

Where the insulation is adhered rather than fastened, ribbon spacing is the equivalent variable and it closes up in the same zones. Adhesives bring constraints fasteners do not: a minimum substrate temperature, a maximum open time, a flatness tolerance on the deck, and a requirement that the board is walked in while the adhesive grabs. A ribbon layout applied to a dusty or damp board is the one attachment method that can look perfect and hold nothing, because there is no torque reading and no plate to inspect — which is the case ASTM E907 exists to settle on a finished roof.

Where the Field Stops Being the Field

Almost every membrane blow-off begins at an edge and almost none in the middle of a field. The mechanism is peel: once wind gets under a free edge it lifts a strip, and the strip acts as a lever on the next row of fasteners, so each row unzips at a fraction of the pressure the assembly resists in tension. Everything at the perimeter is doing one job — denying the wind an edge to start on — so treat the base flashing, the termination bar, the nailer and the edge metal as one detail with four parts.

At a parapet the membrane turns up the wall and terminates on a bar, the last mechanical hold before the flashing becomes a free edge. Manufacturers publish a bar spacing in their own schedule, commonly 300 mm on centre and closer where the wind table calls for it. Count honestly: the run needs one fastener at each end as well as one per interval, splices between bar lengths need their own pair, and corners need fasteners either side of the break rather than one at the mitre. Type follows the wall — masonry anchors into block, self-drillers into steel, a lag into the nailer where the detail lands on timber.

Where there is no parapet, the membrane wraps over the edge onto a nailer under fascia or a gravel stop, and the assembly falls under ANSI/SPRI ES-1, Wind Design Standard for Edge Systems Used with Low Slope Roofing Systems. The model building code adopts ES-1 for perimeter edge securement on low-slope roofs, with the local amendment deciding the edition. A tested edge system is a fascia and a continuous cleat together, sized for the face height in front of them — so a re-roof that thickened the build-up has a taller face than the old metal was tested at, and that metal does not come back on however sound it looks from the ground.

The nailer is the part nobody checks and the part that fails quietly. It has to be anchored for the same uplift the metal above it resists, its fastening has to reach the wall or the deck's supporting member rather than the deck sheet, and its withdrawal capacity in timber is what the whole edge hangs from. Where the roof gained thickness the nailer gained courses, and stacked boards need fastening to each other as well as to the wall: a three-course nailer lagged only through its bottom board is an edge detail waiting to leave in one piece.

Held Down by Weight Instead of Thread

Not everything on this roof resists uplift with a fastener. A ballasted single-ply system holds the sheet down with stone or pavers, pedestal pavers sit loose on the membrane, and a ballasted photovoltaic array is held entirely by its own mass and its friction on the roof surface. ANSI/SPRI RP-4, Wind Design Standard for Ballasted Single-ply Roofing Systems, governs the first of those, and it does what the fastening tables do: it increases the ballast at the perimeter and again at the corners, and it sets a height and an exposure above which ballast is not acceptable at all.

Scour is the failure mode, and it is why the corner increase is not optional. Wind crossing gravel at the edge lifts the smallest stones first, and once a patch is bare the sheet under it is exposed and the surrounding ballast has lost its edge restraint. On a paver roof the equivalent is one lifted paver that becomes a missile with a whole warehouse of run-up. Any loose-laid system needs its perimeter courses heavier, larger or mechanically restrained, and the roof needs a maintenance instruction saying so — the person redistributing ballast around a new curb in five years will not have read RP-4.

A rooftop array adds a question the roofing scope does not usually own. Ballast holding an array down is dead load the deck and joists have to carry, concentrated at the array feet rather than spread, and it lands in the same corner zone where the roof already needed its heaviest attachment. Wind on an array is not the same load case as wind on a bare roof either — ASCE 7 carries its own provisions for rooftop solar panels, and tilt, row spacing and setback from the edge all change the answer. Settle the ballast figure and the structural check together, before anything is craned up.