By The Roofing Brief Team. Last reviewed: August 2026.
Membrane attachment methods compared side by side come down to four options: fully adhered, mechanically attached, ballasted, and induction welded. The choice is set by the wind uplift pressure the roof has to resist and by what the deck can hold, not by installer preference. Each method is rated by a different test standard, each loads the building differently, and each fails in its own way.
The four membrane attachment methods compared
Fully adhered systems bond the membrane to the substrate across its entire surface. Mechanically attached systems fasten the membrane at engineered intervals through plates into the deck. Ballasted systems lay the membrane loose and hold it with stone or pavers. Induction welded systems fuse the membrane to coated plates that were already fastened through the insulation, with no fastener passing through the membrane.
| Method | How it resists uplift | Dead load added | Install speed | Main failure mode |
|---|---|---|---|---|
| Fully adhered | Continuous adhesive bond over the whole surface | Negligible | Slowest, adhesive dwell and weather dependent | Adhesive delamination, peel from a leading edge |
| Mechanically attached | Discrete fasteners and stress plates on an engineered pattern | Negligible | Fastest | Flutter and billowing between fasteners, membrane tearing at plates, fastener backout |
| Ballasted | Gravity, from 10 to 22 lb per sq ft of stone or pavers per ANSI/SPRI RP-4 (2022) | Highest by a wide margin | Fast once the deck is verified | Aggregate scour and blow-off, ponding hidden under stone |
| Induction welded | Welded plates on a grid, fasteners stop below the membrane | Negligible | Comparable to mechanically attached | Weak or missed welds, plate corrosion at the deck |
The dead load column is the one that decides most retrofits. Adhered, mechanically attached and induction welded systems all add a few pounds per square foot. Ballast adds at least 10 lb per sq ft and can reach 22 lb per sq ft under ANSI/SPRI RP-4 paver provisions, which is a structural question before it is a roofing question.
Why attachment method is a wind and deck decision
Wind uplift is a suction load on the top of the membrane, and the attachment method is the only thing transferring that suction back into the structure. The design pressure comes from ASCE 7 and rises sharply in perimeter and corner zones, so the same roof usually needs different attachment densities in different zones. The deck sets the ceiling: a thin gauge steel or lightweight insulating concrete deck limits fastener pullout, and a wood or gypsum deck behaves differently again.
ANSI/FM 4474 (December 2020) makes the interaction explicit in Section 1.2.3: “The performance of a roof assembly depends in part on all components in the roof system and on how they interact.” That is why FM ratings apply to a named assembly, deck gauge and fastener pattern included, and not to a membrane on its own.
How each method is actually tested and rated
Three standards carry the load in North American practice, and they are not interchangeable. ANSI/FM 4474 tests a complete assembly to failure under static pressure. UL 580 classifies an assembly at set pressure levels. ASTM E1592 covers sheet metal panels rather than membranes. The 2021 IBC Section 1504.3.1 accepts FM 4474, UL 580 or UL 1897 for built-up, modified bitumen and single-ply systems.
| Standard | What it covers | How it runs |
|---|---|---|
| ANSI/FM 4474 (2020) | Complete roof assemblies using fasteners, adhesives, hot asphalt, heat welding or self adhesive components, plus standing and lap seam metal | Pressure raised in 15 lb per sq ft increments, each held 60 seconds; the rating is the highest level held for the full 60 seconds |
| UL 580 | Uplift resistance of roof assemblies, including through fastened and standing seam metal panel systems | Classes 15, 30, 60 and 90; static pressure held 5 minutes and oscillating pressure applied at a 10 second frequency for 60 minutes |
| ASTM E1592 | Sheet metal roof and siding panels from 0.012 to 0.050 in. thick, and the anchor to panel attachments | Uniform static air pressure difference across a test chamber |
| ANSI/SPRI RP-4 (2022) | Ballasted single-ply systems, which FM 4474 explicitly does not cover | Prescriptive design tables by wind speed, building height, parapet height and exposure |
For metal panel roofs the pairing is worth getting right. FM 4471 (March 2025) is the examination standard for Class 1 Panel Roofs, covering fire, wind, foot traffic and hail. It does not define its own uplift test: Section 4.3 states that “the wind uplift test procedures shall be in accordance with ANSI FM 4474.” So a Class 1-90 or Class 1-180 panel roof rating is an FM 4471 designation earned through an ANSI/FM 4474 test.
The rating gap nobody mentions
ANSI/FM 4474 Section 1.2.6 says plainly: “This standard is not intended to be used to evaluate loose laid ballasted roof assemblies.” A ballasted roof therefore has no FM uplift number in the same currency as an adhered or mechanically attached assembly. Comparing a “1-90” adhered system against a ballasted roof as if both carried a tested psf value is a category error. Ballast is designed prescriptively under ANSI/SPRI RP-4, referenced by IBC Section 1504.4 in the 2021 edition and renumbered in later editions.
Fully adhered: best load spreading, most surface prep
A fully adhered membrane is bonded across its entire surface, so uplift is resisted continuously instead of at points. FM Approvals defines it as components “bonded to the substrate using a compatible adhesive throughout the entire surface of the roof.” That continuity is why adhered assemblies reach the highest ratings per unit area and why they stay flat in wind rather than lifting between fasteners.
The trade is preparation and weather. FM 4470 (April 2022) applies a derating to the best case: for assemblies with all components fully adhered, the final rating is the maximum pressure held for 60 seconds divided by 0.85, rounded down to the next 15 lb per sq ft increment, capped at 1-990. The standard also requires that all adhesive “maintain full contact between all the surfaces of all components,” with no separation, cracking or peeling permitted.
Adhesive temperature limits are hard limits
Adhesives are the schedule risk. Carlisle’s published data sheet for Aqua Base 120, a water-based bonding adhesive for EPDM, TPO and FleeceBACK membranes, states it “is to be applied when the ambient temperature is 40°F (4°C) and rising” and that it should not be applied if ambient temperature will drop below 32°F (0°C) before the adhesive dries. It also warns against storage below 40°F. Solvent-based options carry their own constraints: Carlisle’s Sure-Seal 90-8-30A data sheet lists a coverage rate of 60 sq ft per gallon of finished surface, applied to both substrate and membrane, and requires the adhesive to dry until tacky without stringing before mating.
That drying window is where adhered work loses days. A crew can fasten in conditions where it cannot bond, which is the practical reason mechanically attached systems dominate late-season commercial schedules.
Mechanically attached: fastest, and the pattern is engineered
Mechanically attached membranes are secured with fasteners and stress plates at defined intervals, usually in the sheet lap so the next sheet covers them. It is the quickest and generally the cheapest single-ply method, and it works on decks where adhesion is unreliable. The catch is that the fastener pattern is an engineered output, not a default spacing, and it tightens in perimeter and corner zones.
ANSI/FM 4474 draws its own line at density. Assemblies with fasteners spaced greater than 2 ft by 4 ft on center, or with a contributory fastener area greater than 8 sq ft per fastener, must go through the larger 12 ft by 24 ft test rather than the small scale 5 ft by 9 ft test. A pattern that looks generous on a shop drawing can change which test the assembly must pass.
Flutter and billowing are permitted behavior, not a defect
The billowing that alarms building owners is written into the acceptance criteria. FM 4470 Section 4.3.1.1.2 allows that “mechanically fastened membranes shall be permitted to separate and deflect from adjacent components at locations where they are not fastened,” and that “tearing of membrane at fastener/stress distributors is allowed up to ultimate failure.” Adhered assemblies get no such exception.
So the failure sequence differs by design. A mechanically attached roof under sustained wind flexes between fastener rows, works the membrane against plate edges, and can fatigue at those points. Repeated flexing also cycles the fastener in the deck, which is where backout shows up on older steel decks, particularly where a fastener landed in a flute rather than on the top flange. Adhered roofs do not flutter; they peel, and once a leading edge lifts the failure propagates.
Ballasted: cheapest, heaviest, and now restricted
A ballasted roof lays the membrane loose and holds it down with washed river stone or concrete pavers. It is the least labor intensive method and it protects the membrane from UV, but it puts real weight on the structure and it is fenced in by ANSI/SPRI RP-4, which IBC Section 1504.4 (2021 edition) requires for ballasted low-slope single-ply systems below 2:12.
| Ballast type per ANSI/SPRI RP-4 (2022) | Minimum quantity | Approximate dead load |
|---|---|---|
| #4 stone, nominal 1.5 in. river bottom, per ASTM D7655/D7655M | 1,000 lb per 100 sq ft | 10 lb per sq ft |
| #2 stone, nominal 2.5 in. river bottom | 1,300 lb per 100 sq ft | 13 lb per sq ft |
| Standard concrete pavers (#4 designs) | Minimum 18 lb per sq ft | 18 lb per sq ft |
| Concrete pavers (#2 designs) | Minimum 22 lb per sq ft | 22 lb per sq ft |
| Interlocking lightweight pavers, beveled, doweled or contoured fit | Minimum 10 lb per sq ft | 10 lb per sq ft |
RP-4 also caps where ballast can go. The standard is limited to roof slopes up to 2 in 12 measured at the top of the membrane. When building height exceeds 150 ft, the roof must be designed by a registered design professional using ASCE 7 practice. Where parapets are less than 1 ft tall, aggregate ballasted systems are limited to buildings under 75 ft. In wind-borne debris regions, which RP-4 identifies as areas within 1 mile of the coastal high water line at ultimate design wind speeds of 130 mph or more, or anywhere at 140 mph or more and in Hawaii, the design goes to a registered design professional.
High wind ballasted roofs are hybrids
The detail most summaries miss is that RP-4’s higher wind designs stop being ballasted at the edges. In a System 3 design, the corner and perimeter zones require “an adhered or mechanically attached roof system designed to withstand the uplift force in accordance with ASCE 7 or the local building code,” installed “with no loose aggregate placed on the membrane.” Where a protective covering is called for, RP-4 requires an adhered membrane in those zones and states that mechanically fastened membrane systems “shall not be used.” Ballast survives in the field of the roof and disappears exactly where uplift is worst.
Induction welded: the hybrid that removed the penetration
Induction welding, sold as RhinoBond by OMG and offered through Sika Sarnafil and other membrane manufacturers, fastens coated plates through the insulation to the deck first, then heats each plate from above the membrane so the membrane fuses to the plate. No fastener passes through the membrane. Sika’s published description of the system calls for holding a weighted magnet on each welded plate for a minimum of 60 seconds to set the bond.
The performance case is a grid instead of rows. Sika states FM approval “up to 210 PSF” for the system, and that for a 1-90 psf rated roof the method uses “10 fewer fasteners and plates for every square installed when compared to a typical in-seam, mechanically-attached system using a 10-foot wide membrane fastened 12 inches on center.” Distributing plates on a grid also reduces the sheet flutter that mechanically attached systems accept by design.
Two practical limits apply. Weld quality is invisible from above without pull testing, so quality control shifts from a visible seam to a documented process, and the plates must be the manufacturer’s coated plates, which rules out mixing hardware across systems. The method is also the usual answer for light gauge metal retrofits where panel pullout values are too low for conventional in-seam fastening.
Deck compatibility and where each method gets excluded
Deck type narrows the field before wind does. Structural concrete takes adhesive well and takes fasteners poorly without drilling. Steel deck takes fasteners well and takes adhesive only through an adhered cover board. Lightweight insulating concrete and cementitious wood fiber need specific fasteners, and ANSI/FM 4474 sets minimum deck thickness conditions of its own: 2 in. for structural concrete, lightweight insulating concrete, cementitious wood fiber or gypsum decks, and 0.75 in. for fiber reinforced cementitious decks.
| Deck | Fully adhered | Mechanically attached | Ballasted | Induction welded |
|---|---|---|---|---|
| Steel | Yes, over an adhered cover board | Best fit | Only if capacity verified | Best fit |
| Structural concrete | Best fit | Requires drilled anchors | Often viable, capacity permitting | Requires drilled anchors |
| Lightweight insulating concrete | Limited, adhesion varies | Yes, with specified fasteners | Rarely, added load | Yes, with specified fasteners |
| Wood | Yes | Yes | Only if capacity verified | Yes |
| Light gauge metal panel retrofit | Difficult | Low pullout values | No | Common solution |
Exclusions stack up fastest for ballast: slopes above 2 in 12, buildings above 150 ft without engineered design, low parapets on tall buildings, wind-borne debris regions, and any structure whose framing was not designed for an extra 10 to 22 lb per sq ft. That combination, plus the difficulty of finding leaks under stone, is why ballasted work has become uncommon on new construction while remaining in service on a large installed base.
How to select the attachment method
- Calculate component and cladding uplift pressures from ASCE 7 for the field, perimeter and corner zones, using the building’s actual height, exposure and risk category.
- Confirm the deck type, thickness and condition, and get fastener pullout values for the actual deck rather than a generic table.
- Ask the structural engineer whether the frame can carry ballast, in combination with all other design loads, before ballast stays on the option list.
- Match candidate assemblies to a manufacturer system approval that covers your deck, insulation, cover board and fastener pattern as an assembly, not a membrane in isolation.
- Check the code path: FM 4474, UL 580 or UL 1897 for membrane systems under 2021 IBC Section 1504.3.1, ASTM E1592 or FM 4474 for structural metal panels, ANSI/SPRI RP-4 for ballast.
- Apply the schedule filter last. If the work runs into weather below 40°F, adhered assemblies may lose their adhesive window and a fastened or induction welded system is the realistic choice.
Repairability, the cost nobody quotes
Adhered roofs are the easiest to troubleshoot because water cannot travel far under a bonded membrane, so a leak generally sits near the breach. Mechanically attached roofs let water run between fastener rows, and the entry point can be many feet from the interior stain. Ballasted roofs require moving stone to find anything, which turns a small repair into a materials handling job. Induction welded roofs read like adhered roofs for leak tracing since the membrane is bonded at every plate, though a failed weld is only found by pulling on it.
Those differences also shape recover and replacement economics. For the wider material tradeoffs beneath the attachment decision, see how built-up roofing compares with single-ply, and how EPDM and TPO differ on cost and lifespan. Seam integrity carries the other half of the wind performance question, covered in our guide to TPO seam welding temperature and probe testing. System-level approvals are published per manufacturer, and Versico’s TPO, EPDM and PVC lines illustrate how one supplier organizes adhered, mechanically attached and induction welded options. Our commercial membrane market share report tracks how the installed base is shifting.
Frequently asked questions
Which membrane attachment method has the best wind uplift resistance?
Per unit area, fully adhered systems resist uplift best because the bond is continuous rather than concentrated at fasteners. Induction welded systems can post very high tested numbers, with Sika publishing FM approval up to 210 psf for RhinoBond assemblies. Ratings are assembly specific, so the deck, insulation and fastener pattern matter as much as the attachment concept itself.
Why is a ballasted roof not given an FM uplift rating?
ANSI/FM 4474 states in Section 1.2.6 that it “is not intended to be used to evaluate loose laid ballasted roof assemblies.” Ballasted single-ply systems are instead designed prescriptively under ANSI/SPRI RP-4, which the IBC references for ballasted low-slope roofs below 2:12. That means ballasted and adhered systems cannot be compared on a single tested psf figure.
How much weight does ballast add to a roof?
ANSI/SPRI RP-4 (2022) sets minimums of 1,000 lb per 100 sq ft for #4 stone and 1,300 lb per 100 sq ft for #2 stone, which work out to about 10 and 13 lb per sq ft. Paver ballast minimums run from 10 lb per sq ft for approved interlocking lightweight pavers to 22 lb per sq ft for concrete pavers in #2 designs. A structural engineer should verify capacity first.
Is membrane flutter on a mechanically attached roof a defect?
Not by itself. FM 4470 allows that mechanically fastened membranes may “separate and deflect from adjacent components at locations where they are not fastened,” and permits membrane tearing at fasteners and stress plates up to ultimate failure during testing. Visible billowing in wind is expected behavior for the method. Persistent large-scale flutter, loose plates or backed-out fasteners are worth inspecting.
What temperature is too cold to install a fully adhered membrane?
It depends on the adhesive, and the manufacturer’s data sheet governs. Carlisle’s Aqua Base 120 water-based bonding adhesive is specified for application when ambient temperature is 40°F and rising, and should not be applied if the temperature will fall below 32°F before it dries. Cold also lengthens drying time, so a crew may lose most of a working day waiting for tack.
Does FM 4471 replace FM 4474 for metal panel roofs?
No. FM 4471 is the examination standard for Class 1 Panel Roofs, covering fire, wind uplift, foot traffic and hail resistance together. For the uplift portion it directs that “the wind uplift test procedures shall be in accordance with ANSI FM 4474.” A Class 1-90 or Class 1-180 panel roof designation is an FM 4471 classification earned through ANSI/FM 4474 testing.
Can you mix attachment methods on one roof?
Yes, and high wind designs often require it. ANSI/SPRI RP-4 System 3 designs call for an adhered or mechanically attached system in corner and perimeter zones with no loose aggregate placed on the membrane, leaving ballast only in the field. Where a protective covering is required in those zones, RP-4 specifies an adhered membrane and states that mechanically fastened systems shall not be used.
Sources: ANSI/FM 4474 (December 2020); FM Approvals Standard 4470 (April 2022); FM Approvals Standard 4471 (March 2025); ANSI/SPRI RP-4 (2022); 2021 International Building Code Sections 1504.3 and 1504.4; UL 580; ASTM E1592-05(2017); ASTM D7655/D7655M; Carlisle SynTec published product data sheets for Aqua Base 120 and Sure-Seal 90-8-30A; Sika Sarnafil published RhinoBond system documentation. Wind design pressures should be calculated from the ASCE 7 edition adopted by your jurisdiction, and code section numbering has shifted between IBC editions, so verify against the adopted edition.