Roofing
Single-ply roof attachment methods and wind uplift design
How a single-ply membrane and its insulation get held to the deck against wind uplift: mechanically attached, fully adhered, induction-welded, and ballasted, matched to the FM and ASCE 7 wind design.
Direct answer
Single-ply attachment is how the membrane and insulation are held to the deck against wind uplift. The four methods are mechanically attached, fully adhered, induction-welded, and ballasted. Each carries a tested uplift rating that must match the design wind load. FM Global, ASCE 7, and the manufacturer's listing control the assembly.
Key takeaways
- The four single-ply attachment methods are mechanically attached, fully adhered, induction-welded, and ballasted, each carrying a tested uplift rating that must match the design wind load.
- Wind suction is highest at corners, often 2 to 3 times the field pressure, so corners and perimeter get tighter fastening or more adhesive than the field.
- FM Approvals rates assemblies by tested resistance (1-60 held 60 psf, 1-90 held 90, 1-120 held 120) and applies a safety factor of 2.
- Ballasted single-ply per ANSI/SPRI RP-4 caps at a 2 in 12 slope, commonly 10 psf of stone in the field and about 20 psf at the perimeter.
- Pull-test fasteners in the actual deck per FM DS 1-52 to back out the fastener count; never trust a catalog pullout, especially on recovers or aged decks.
Single-ply attachment, and why it keeps the roof on
Single-ply attachment is the method that holds the membrane and the insulation to the roof deck so the wind cannot pull them off. It is a separate decision from which membrane you pick. You choose TPO, PVC, or EPDM for the chemistry and the climate, then you choose how that sheet gets anchored: mechanically attached, fully adhered, induction-welded, or ballasted. The membrane selection guide covers the first call. This one covers the second.
The attachment is what actually keeps the roof on the building. Get the membrane right, under-attach it, and you still have a roof that flies off in a storm. The method also drives the cost and the performance more than most owners expect. Mechanically attached is the cheapest and fastest. Fully adhered costs more and rides out wind better. Ballasted is cheap where the building can carry the weight and the slope allows it. The pick is a tradeoff between budget, deck, slope, and the design wind, not a default.
The piece people skip is that attachment is a system, not a single layer. The insulation has to be held to the deck, the cover board to the insulation, and the membrane to whatever is under it. A roof blows off at its weakest connection. Tighten the membrane and leave the insulation loosely fastened and the whole package peels at the insulation line.
Why does wind lift a roof instead of pushing it down?
Wind does not push a low-slope roof down. It pulls it up. As air moves across the building it speeds up over the roof and creates negative pressure, a suction, on the top surface. At the same time air pressure can build inside the building and push up from below. The two add, and the net force tries to lift the assembly off the deck. That is wind uplift, and it is measured in pounds per square foot.
The suction is not even across the roof. It is highest at the corners, next highest along the perimeter, and lowest in the broad field in the middle. Corner pressures often run 2 to 3 times the field pressure. That is why the attachment is not uniform. The corners and the perimeter get tighter fastening or more adhesive than the field, sized to the higher load in those zones.
The edge is where uplift turns into a failure. Once the wind gets a lip of edge metal or a loose corner to work on, it peels the membrane back and the field unzips behind it. The roof edge and coping wind design is its own subject, covered in the edge metal guide. The attachment and the edge work together. A perfectly fastened field still fails if the perimeter edge lets go first.
Mechanically attached: fasteners and plates at the seams
A mechanically attached system fastens the membrane to the deck with screws and round metal stress plates, set in rows along the seams. The next sheet laps over the row of plates and the seam is heat-welded shut, so the fasteners end up hidden under the weld. The fasteners carry the uplift load straight from the membrane into the deck at those discrete points. Between the rows, the sheet is not held down at all.
This is the cheapest and fastest method, which is why it is the default on big-box and warehouse roofs over a steel deck. It installs across a wide temperature range and goes down quickly. The cost lives in the fastener count, and the wind design sets that count.
The catch is billowing. Because the membrane is only held at the seam rows, wind pulls air into the assembly through laps and deck flutes, and the field of the sheet lifts and flutters between the fasteners. You can watch it breathe on a windy day. Mild billowing is normal in this system. Heavy billowing is a problem. It fatigues the membrane, works the seams, and can pull the plates or tear the sheet at the fastener. The fix for a windy site is a tighter fastener pattern, half-width sheets at the perimeter and corners, or a different method. Plan for the field to move, then bound how much.
Fully adhered: the membrane glued flat
A fully adhered system glues the membrane to the substrate across its whole area with a bonding adhesive, so there is no fluttering field and no fastener through the membrane. The adhesive is solvent-based, water-based, or a low-rise spray foam, rolled or sprayed onto the cover board, and the sheet is broomed into it. The insulation and cover board below are still fastened or adhered to the deck. The membrane itself is bonded, not pierced.
Adhered systems give the best wind uplift performance of the common methods, because the load spreads across the entire bonded area instead of concentrating at fastener points. There is no billowing, the seams see less working, and the roof is quiet. It is the choice when the design wind is high, when the deck cannot take a dense fastener pattern, or when the owner wants the longest warranty.
It costs more and goes slower. The adhesive has a temperature and dew-point window, the substrate has to be clean and dry, and the bond needs time to build. Manufacturer data points to bond strength climbing over roughly the first month after install. On a re-roof you rarely get a textbook-clean substrate, so the schedule gets built around the weather, not the drawing. Glue the sheet to a damp or dusty cover board and you own the blow-off, because the bond that looked fine at install never reached strength.
Ballasted: weight holds the membrane down
A ballasted system lays the membrane loose over the substrate and holds it down with weight: smooth river rock, crushed stone, or concrete pavers spread over the top. Nothing penetrates the membrane in the field. The weight resists the uplift. It is the lowest material cost of the methods, and it isolates the membrane from the structure, which can help with thermal movement and sound.
The constraints are real and they rule it out on a lot of buildings. ANSI/SPRI RP-4 is the wind design standard for ballasted single-ply, and it caps the method at a roof slope of 2 in 12, about 10 degrees. Steeper than that and rock migrates, and a registered design professional has to sign off. The standard sets ballast weight by zone: commonly on the order of 10 pounds per square foot of stone in the field and about 20 pounds per square foot at the perimeter and around large penetrations, with interlocking pavers carrying their own minimum weight. The structure has to carry that dead load, which is the first thing to check.
Wind scour is the failure to watch. In a high wind the stone at the corners and perimeter can be scoured away, uncovering the membrane and letting the uplift start. Building height, parapet height, exposure, and ballast size all feed the RP-4 design. On tall or exposed buildings ballast often does not pencil out, and you move to larger stone at the perimeter, pavers, or a fastened method.
Induction-welded (RhinoBond-type): plates welded to the membrane
The induction-welded system, known in the field by the RhinoBond name from OMG and used by Sika and others, is a hybrid. Specially coated metal plates are screwed through the insulation into the deck on a grid, the same plates that hold the insulation down. Then the membrane is laid over the plates and an induction tool fuses the sheet to the coating on each plate from above. No fastener goes through the membrane.
It combines the strengths of the other two. There is no penetration of the membrane like an adhered roof, so the water risk at the fastener is gone, and the sheet is held to a grid of points so it does not flutter like a loose mechanically attached field. It uses fewer plates and fasteners than a comparable mechanically attached pattern for the same uplift, because each plate holds both the insulation and the membrane and the load spreads off the welded disc.
The plates are typically 3 in coated steel. The tool heats the plate, then a weighted magnet sits on the spot for about a minute while the bond sets. The weld has to actually take. A plate the tool missed or a spot welded cold is a hold-down that is not holding, and you find it with a pull test, not by eye. It has become the fastest-growing attachment method in commercial single-ply for these reasons. It costs more than plain mechanical attachment and has less weather drama than full adhesion in cold or damp conditions.
How is the fastening pattern set for the wind zones?
The fastening pattern is not one spacing across the roof. It is set zone by zone to the uplift in each zone, which means the field gets the loosest pattern, the perimeter gets tighter, and the corners get tightest. The zones and the pressures come from ASCE 7, and the manufacturer's listed assembly tells you the fastener spacing or the adhesive that holds each zone's load.
In a mechanically attached roof the common move is to narrow the sheets and add fastener rows toward the edge. Full-width sheets in the field, half-width or narrower sheets at the perimeter and corners, so the seam rows, and the fasteners in them, come closer together exactly where the suction is worst. Some designs add intermediate fastener rows instead of cutting the sheets. Either way the corner can carry 2 to 3 times the field fastener density.
In an adhered roof the same logic uses more adhesive coverage or a stronger bond at the perimeter and corners, and the insulation fastening under it tightens in those zones too. The mistake that blows roofs off is a uniform field pattern carried all the way to the edge. The field rating is the lowest rating on the roof, and the corners are where the wind starts. Enhance the perimeter and the corners or the design is wrong no matter how good the field number looks.
How is roof wind uplift designed?
Roof wind uplift design is a two-step match: calculate the uplift the building will see, then specify an assembly tested to beat it with a margin. The load comes from ASCE 7, which turns the site wind speed, the building height and exposure, and the roof geometry into a design pressure in pounds per square foot for each zone. The resistance comes from a tested assembly, either an FM Approvals rating or the membrane manufacturer's listed system.
FM Global is the common path on insured commercial work. FM Approvals rates an assembly by the pressure it resisted in test. A 1-60 rating held 60 pounds per square foot, 1-90 held 90, 1-120 held 120. FM applies a safety factor of 2, so a 1-60 assembly is recommended where the design uplift is about 30 pounds per square foot. The load side is set up in FM data sheets, with DS 1-28 covering design wind loads and DS 1-29 covering deck securement and the above-deck components. ANSI/SPRI also publishes wind design standards, RP-4 for ballasted and a wind design standard practice for membrane systems.
The rule is simple and it is where roofs are lost. The tested rating has to cover the calculated load, zone by zone, with the safety factor. A field rating that beats the field load means nothing if the corner assembly was never rated for the corner load. The manufacturer's listing and FM control the specific numbers, and the adopted building code, which references ASCE 7, sets the floor.
The deck and the fastener have to match
The deck decides the fastener, and the fastener's pullout decides how many you need. A screw that holds beautifully in 22-gauge steel does nothing in a worn lightweight concrete deck. Steel deck takes a self-drilling fastener and gives consistent pullout. Structural concrete takes a different anchor and usually holds well. Wood and plywood take a coarse-thread fastener, but thin or old plywood and OSB give up withdrawal fast, especially under 3/4 in where there is not enough thread engagement and the laminations have voids.
Lightweight insulating concrete and gypsum decks are the tricky ones. Their density runs all over the map depending on mix and age, so the pullout is not a catalog number. A wet gypsum deck loses withdrawal resistance, and a fastener that spins in soft fill is not holding anything. The only honest way to know what a questionable deck gives is to pull-test fasteners in the actual deck, not trust a table.
That field pull test is the input to the whole pattern. You measure the withdrawal a single fastener achieves in the real deck, apply the safety factor, and back out how many fasteners per square the wind design needs. FM DS 1-52 covers field verification of roof wind uplift resistance for this reason. Skip the pull test on an existing deck and you are guessing at the one number the entire attachment depends on.
Where the seam and the attachment meet
In a mechanically attached roof the fastener lives inside the seam. The row of plates and screws sits along the edge of one sheet, the next sheet laps over them, and the heat weld closes the lap over the top of the fasteners. So the seam is doing two jobs at once. It is the watertight joint and it is the cover over the load-bearing fastener row. A weak weld over a fastener row is both a leak and a hold-down failure waiting to happen.
That is why the weld quality matters more in a mechanically attached system than people credit. The uplift load pulls on the fastener, the fastener pulls on the membrane right at the weld edge, and a cold or narrow weld peels under that working load. The seam welding and probe-testing is its own subject, but the attachment is the reason the seam over a fastener row is the highest-stress seam on the roof.
In a fully adhered roof the seams are not carrying the field fasteners, so the seam and the attachment are more independent, but the seam still has to be sound because it is the water path. Whichever system, the seam gets checked the same way: a probe along the weld and a destructive cut on a test seam to confirm the weld, not the eye.
Attaching over an existing roof
A recover, or overlay, goes over the existing roof instead of tearing it off, and the attachment has to reach through the new layers into something that actually holds. The common approach is a recover board over the old membrane, then the new single-ply attached through the board. The fastener still has to land in the original structural deck to get its pullout, because the old membrane and the wet insulation under it hold nothing.
The trap is the deck you cannot see. On a recover you are fastening into a deck that has been under a roof for 20 years, and it may be corroded steel, soft wet gypsum, or deteriorated lightweight concrete. The pull test matters even more here than on new work, because the assumption that the deck still gives its original withdrawal is exactly the assumption that fails. Pull-test through the recover assembly into the real deck before the pattern is set.
A recover also adds load and height the wind design has to account for, and it raises the roof relative to the edge metal and the parapet, which ties back to the edge detail. If the recover lifts the field above the existing edge, the edge no longer caps it, and the perimeter is the first thing the wind finds.
Cost versus wind performance by method
The four methods line up in a rough order on both cost and wind performance, and the two do not move together. Mechanically attached is the cheapest and the most prone to billow. Fully adhered costs the most and rides out wind the best. Induction-welded sits in between, buying adhered-like field behavior without the adhesive's weather sensitivity. Ballasted is cheap on materials but limited by slope, structure, and scour. The right pick matches the method to the design wind, the deck, the slope, and the budget, in that order, not to habit.
| Method | Relative cost | Wind behavior | Where it fits |
|---|---|---|---|
| Mechanically attached | Lowest | Field flutters and billows, load at fastener points | Low to moderate wind, sound steel deck, budget jobs |
| Fully adhered | Highest | No flutter, best uplift, load spread across the bond | High wind, decks that cannot take dense fasteners, long warranty |
| Induction-welded | Moderate | No membrane penetration, no flutter, fewer fasteners | High wind, cold or damp weather, hybrid need |
| Ballasted | Low materials | Weight holds it, corners can scour | Slope under 2:12, structure carries the weight, low to moderate exposure |
What changes for high-wind and coastal roofs?
On a high-wind or coastal building the uplift loads climb fast and the cheap default stops working. Corner and perimeter pressures that were a nuisance inland become the governing load, and a plain mechanically attached field that flutters is the wrong call. Adhered and induction-welded systems move to the front because they hold the field flat and spread the load, and the perimeter and corner enhancement gets aggressive.
In hurricane-prone regions the design also leans on stricter testing and approvals. Florida product approval and Miami-Dade notices of acceptance, and FM's higher ratings, are how an assembly proves it survives the design wind. The fastener spacing tightens, often to 4 in at panel edges in the worst zones where a milder site might use 6 in. The edge metal has to meet the matching wind rating too, which is the edge guide's subject.
The blunt version: in a coastal corner zone the roof either has a tested assembly rated for the real load with the safety factor, or it is a future insurance claim. This is not the place to value-engineer the fastener count or skip the perimeter enhancement. The wind finds the corner first, every time, and the corner is where under-design shows up as a peeled roof.
Large roofs: data centers and distribution buildings
On a very large roof, a data center, a distribution center, a warehouse running into the millions of square feet, the attachment decision scales into real money and real risk. The field is enormous, so the fastener count or the adhesive quantity in the field drives the budget, and small changes in the pattern multiply across acres. At the same time the consequence of a blow-off is severe, because the building under it is full of expensive and sensitive contents.
The field of a big roof is mostly low-pressure field zone, which tempts a thin pattern, but the perimeter and corners are still where it fails, and on a long building the perimeter is a lot of linear footage. Induction-welded systems are common here because they cut the fastener count in the field while holding the sheet flat, and adhered systems show up where the wind or the owner's risk tolerance demands it.
The other large-roof reality is the deck survey. Across acres of deck the pullout is not uniform, especially on a recover, so the pull testing has to sample the deck in enough spots to catch the soft areas instead of taking one lucky reading. One weak zone on a huge roof is the zone that lets go.
How is the attachment checked before the roof is signed off?
The attachment is verified by test, not by appearance, and the checks differ by method. On any fastened system the first check is the fastener pull test, done on the actual deck before the pattern is finalized and spot-checked during the work, because the deck withdrawal is the number the whole design rests on. The installed pattern is then counted against the listed assembly: right fastener, right plate, right spacing in each zone.
On a mechanically attached or induction-welded roof the seam weld is probed along its length with a rounded tool and confirmed with destructive test cuts on sample seams, because the seam over a fastener row carries load as well as water. The seam QA is its own subject. On an induction-welded roof the plates also get pull-tested to confirm the welds actually took, since a missed or cold weld looks identical to a good one from above.
On an adhered roof the inspector checks for full contact and no blisters or fishmouths, and confirms the substrate was clean and dry and within the adhesive's temperature window when it went down. Across all of them the perimeter and corner enhancement gets verified specifically, because that is the zone that fails and the zone most often shorted in the field. The roof that passes is the one where the corner pattern matches the corner design, not just the field.
What to document
The attachment record is what proves the roof was built to the wind design, and what a reviewer reads when a corner lifts in the first big storm. Capture it by zone, because the zones are where the design varies and where failures concentrate. For each zone, record the method, the fastener or adhesive used, the pattern or coverage, and the tested uplift rating that zone was built to, along with the field pull-test results that justified the fastener count. The values below are illustrative; the listed assembly and the project documents set the real numbers.
| Zone | Method | Fastener or adhesive | Pattern or coverage | Uplift rating |
|---|---|---|---|---|
| Field | Mechanically attached | Fastener and 3 in plate per listing | Seam rows per listed spacing | FM rating per design load |
| Perimeter | Mechanically attached | Same fastener and plate | Narrower sheets, tighter rows | Enhanced for the zone load |
| Corner | Mechanically attached | Same fastener and plate | Tightest rows or half sheets | Highest zone rating |
| Deck verification | All zones | Field pull-test value | Safety factor applied | Listed assembly number |
Common mistakes
- Under-attaching the field, or carrying a field pattern all the way to the edge with no perimeter or corner enhancement.
- Specifying a field uplift rating that beats the field load while the corner assembly was never rated for the corner load.
- Using the wrong fastener for the deck, or trusting a catalog pullout instead of pull-testing the actual deck.
- Leaving a mechanically attached field to billow heavily instead of tightening the pattern or changing the method.
- Ballasting a roof steeper than 2 in 12, or where the structure cannot carry the load or the corners will scour.
- Gluing an adhered membrane to a damp, dusty, or cold substrate outside the adhesive's temperature and dew-point window.
- Skipping the seam probe and destructive cuts over fastener rows, or not pull-testing induction-welded plates.
- Matching no tested assembly to the calculated ASCE 7 load with the FM safety factor.
Field checklist
Want this checklist to run itself on every job — with photo proof and a signed record crews can hand the customer? That's FieldOS.
Standards and references
FM Global is the common framework on insured commercial roofs. The load side sits in Data Sheet 1-28 for design wind loads, built on ASCE 7, and Data Sheet 1-29 covers roof deck securement and the above-deck components. FM Approvals lists assemblies by their tested resistance, the 1-60, 1-90, and 1-120 windstorm classifications in pounds per square foot, with a safety factor of 2 between the tested number and the recommended design load. Data Sheet 1-52 covers field verification of wind uplift resistance, the basis for pull testing.
ASCE 7, Minimum Design Loads, is where the uplift pressure and the field, perimeter, and corner zones come from, and the building code references it for the design load. ANSI/SPRI publishes wind design standards for roofing, including RP-4 for ballasted single-ply and a wind design standard practice for membrane systems. NRCA gives installation guidance and detail practice.
Above all of these sits the membrane manufacturer's listed system and warranty. The manufacturer tests specific assemblies and will only warrant the roof if it is built to a listed configuration for the wind zone, so the listing often controls the exact fastener, plate, spacing, and adhesive. The exact data sheet and standard numbers and editions change, so confirm them against the current FM data sheets, the manufacturer's current approval, and the adopted code before citing them on a submittal.
Units, terms, and conversions
Wind uplift design carries a few units and names that read differently across a spec, an FM letter, and a manufacturer's approval. The same idea shows up as pounds per square foot of pressure and as the FM classification number that stands for it, so the numbers have to be read in the right system before they are compared.
- Wind uplift
- The net upward suction force on the roof, in pounds per square foot (psf)
- FM 1-60 / 1-90 / 1-120
- FM Approvals windstorm classification: the psf the assembly resisted in test, before the safety factor of 2
- ASCE 7
- The standard that sets design wind loads and the field, perimeter, and corner zones
- Field / perimeter / corner
- Roof zones by uplift, lowest in the field and highest at the corners
- Pullout / withdrawal
- The force needed to pull a fastener out of the deck, measured by field pull test
- Stress plate
- The round metal plate under a fastener that spreads the load into the membrane
- Ballast
- Stone or pavers laid over a loose membrane to hold it with weight, designed per ANSI/SPRI RP-4
FAQ
Mechanically attached vs fully adhered roof: which is better for wind?
Fully adhered handles wind better because the load spreads across the whole bonded area instead of concentrating at fasteners, and the field cannot billow. Mechanically attached is cheaper and faster but flutters and works the seams. For high-wind sites, adhered or induction-welded usually wins; the project's design uplift and deck control the call.
What is a ballasted roof?
A ballasted roof is a loose-laid single-ply membrane held down by weight: smooth stone, crushed rock, or concrete pavers spread over the top, with nothing penetrating the field. It is low material cost but limited to slopes under 2 in 12, needs structure to carry the load, and is designed for wind per ANSI/SPRI RP-4.
Why do mechanically attached roofs billow?
Mechanically attached membranes are only fastened in rows at the seams, so between rows the sheet is free. Wind pulls air into the assembly through laps and deck flutes and the field lifts and flutters. Mild billowing is normal; heavy billowing fatigues the membrane and can pull fasteners, and it calls for a tighter pattern or another method.
How is roof wind uplift designed?
Calculate the uplift from ASCE 7 using the site wind speed, building height, exposure, and roof geometry, which gives a design pressure in psf by zone. Then specify an assembly tested to beat it, an FM rating or the manufacturer's listed system, with the safety factor. The tested rating has to cover the load in every zone.
What is induction welding or RhinoBond on a roof?
Induction welding, known as RhinoBond from OMG and used by Sika and others, screws coated plates through the insulation into the deck, then fuses the membrane to the plates from above with an induction tool. No fastener goes through the membrane, the field does not flutter, and it uses fewer plates than a comparable mechanically attached pattern.
What does an FM 1-90 rating mean?
An FM 1-90 rating means the roof assembly resisted 90 pounds per square foot of uplift in the FM Approvals test. FM applies a safety factor of 2, so a 1-90 assembly is recommended where the calculated design uplift is about 45 psf. The rating is a pressure class, not a wind speed, and it applies zone by zone.
Do fasteners penetrate the membrane on every system?
No. Mechanically attached systems put fasteners through the membrane at the seams, hidden under the weld. Fully adhered and induction-welded systems do not penetrate the membrane: the adhered sheet is glued and the induction-welded sheet is fused to plates from above. Ballasted membranes are loose in the field with no penetrations there either.
Why does the corner of a roof need more fasteners?
Wind suction is highest at the corners, often 2 to 3 times the field pressure, next highest at the perimeter, and lowest in the field. The attachment has to match each zone, so corners and perimeters get tighter fastening, narrower sheets, or more adhesive. A uniform field pattern carried to the edge is what blows roofs off.
Do I need to pull-test fasteners before reroofing?
Yes, especially on a recover or an old deck. Fastener pullout depends on the actual deck, and lightweight concrete, gypsum, and aged steel or thin plywood vary widely, so a catalog number is a guess. Pull-test in the real deck, apply the safety factor, and back out the fastener count. FM DS 1-52 covers this field verification.
People also ask
Codes cited in this guide
This guide is written and reviewed against the published standards below. Always confirm the current adopted edition with the authority having jurisdiction.