EPDM roofing systems
Low-slope · commercial and industrial
The oldest continuous field record in single-ply belongs to a sheet that almost never fails. What failed, and what was fixed, happened at the lap.
What should an owner actually judge an EPDM roof on?
EPDM is a cured synthetic rubber sheet with the longest continuous field record of any single-ply membrane, and that record is mostly a record of seams. Federal research traced the early failures to the bond at the lap, not to the rubber. Judge an EPDM proposal on seam method, attachment, and termination detail. Colour is an energy and moisture decision, not a durability claim.
The short versionSection link
EPDM is a thermoset: a vulcanised rubber that cannot be re-melted. Everything distinctive about the system follows from that one fact.
- Material class
- Thermoset rubber sheetCured, so heat will not re-fuse it. Laps are bonded with tape or adhesive, never welded.
- Product standard
- ASTM D4637/D4637M-15(2021)e1Type I non-reinforced, Type II internally reinforced with fabric or scrim, Type III fabric-backed.
- What the standard covers
- The sheet onlyIt sets minimum sheet quality and explicitly excludes in-place system considerations such as fire resistance and uplift.
- Colour options
- Black or whiteA reflectance decision with an energy and a moisture consequence in both directions. Not a durability ranking.
- Attachment
- Ballasted, mechanically attached, or fully adheredA wind design for this building, produced by a qualified designer against the adopted code and the AHJ.
- Service life
- No figure is published hereThe variables that decide it — seam method, attachment, detail count, traffic, maintenance — are on this page instead.
Where EPDM fits, and where it does notSection link
These are conditions, not preferences. Each one is a fact about a building rather than a fact about a sheet.
Best when
- The roof is a large, simple field with few penetrations, so the number of laps and terminations is low.
- The building has no kitchen exhaust, no process exhaust carrying oils, and no hydrocarbon exposure on the roof.
- The owner intends to repair rather than replace at the first leak, and wants a system a competent crew can patch without a hot-air welder.
- The deck and structure have been evaluated and can carry the attachment method the wind design calls for — including, where it is chosen, the permanent dead load of ballast.
- Someone will actually inspect and maintain it. EPDM rewards a maintenance regime more than it rewards a thicker sheet.
Think twice if
- The roof sees animal fats, vegetable oils, or petroleum products — EPDM is documented as vulnerable to all three, and this is the exposure that most often rules it out.
- The building carries heavy, unscheduled rooftop service traffic and no walkway system is in the budget.
- The energy model or the adopted energy code is pushing toward a reflective surface, and the design has not been re-checked for the moisture consequence of removing the heat that used to dry the assembly downward.
- The proposal is priced on the sheet and silent on seam method, primer, roller procedure, and termination detail. That proposal is not about EPDM at all.
- Nobody local is fluent in EPDM detailing. Repairability is only an asset where someone can perform the repair correctly.
What changes the answer
- The exposure map of the roof: what the building itself puts on its own membrane decides more than the polymer does.
- Whether the deck and structure have been evaluated — that gates ballast entirely and constrains mechanical attachment.
- The climate zone and the adopted energy code, which may not leave the colour decision open.
- The detail count. A roof that is mostly curbs, drains and pipe supports is a flashing project wearing a membrane, and the field sheet stops being the deciding variable.
- Whether the owner is buying a roof or buying a maintenance programme. On EPDM those are close to the same purchase.
The sheet is the part that does not failSection link
EPDM has been on American roofs since the 1970s. Almost none of the trouble in that record is located in the rubber.
EPDM — ethylene propylene diene terpolymer — is a thermoset. It is vulcanised during manufacture, which means the polymer chains are chemically cross-linked and the sheet cannot be melted back into a liquid. That is why an EPDM roof is durable in ways that are easy to take for granted: it does not depend on plasticisers staying in the sheet, and its resistance to ozone, heat aging and weathering is written into the product standard, ASTM D4637, alongside brittleness point, tear strength, and both dynamic and static puncture resistance.
It is also why an EPDM lap cannot be welded. Two thermoplastic sheets under hot air become one material. Two cured rubber sheets stay two materials, and something has to be put between them. That something — and the surface it is asked to stick to — is where the history of this system happened.
The five conditions in the drawing
- The sheet face as shipped. EPDM leaves the factory carrying a talc-like release agent so the roll does not bond to itself. NIST’s scanning-electron work on seams cut out of real roofs found that residue on the bonding face, and recorded plainly that it was not visible to the unaided eye. An installer cannot see the problem he is about to bond over.
- Cleaner and primer. This is the step that converts an as-shipped surface into a bond face. In NIST’s laboratory work supporting the Fort Belvoir investigation, seam strength fell by about 40 per cent when the rubber surfaces were not cleaned at all. The same report found the talc-like contamination still present on some specimens that had been cleaned with the recommended wash solution following the recommended technique, and concluded that a means of assuring the quality of cleaned rubber was needed.
- The bonding layer. Historically a liquid contact adhesive applied on the roof; from the 1980s onward, progressively displaced by a factory-made tape. This substitution is the single largest change in the system’s history and it is covered in its own section below.
- Pressure. Tape does not bond because it is sticky; it bonds because it is rolled. NIST’s consortium experiments treated applied pressure as a designed variable alongside surface condition, primer, temperature and dwell time, testing at 0.2 and 2 megapascals. On a roof, that variable is a person with a roller and a schedule.
- Lap sealant at the exposed edge. The seam edge is where water, dirt and ultraviolet light reach the bond line. It is also the cheapest thing to leave out, and the last thing an owner can check from a photograph.
Notice what is absent from that list: the rubber. An owner reading a proposal that specifies “60-mil EPDM” has been told the least consequential fact about the roof being sold. The mil thickness of the sheet is real, and it matters for puncture and hail, but it is not where this system’s record was written.
What the federal seam research actually foundSection link
Between 1988 and 1998 the National Institute of Standards and Technology, working with the Army and with an industry consortium, took EPDM seams apart. It is the best-documented failure investigation in single-ply roofing, and almost nobody selling a roof mentions it.
The sequence matters, and NIST’s own retrospective on this work sets it out plainly. EPDM arrived in the United States in the 1970s and won market share fast: by the mid-1980s it had captured about a third of the total low-slope roofing market. Also by the mid-1980s, surveys conducted by the National Roofing Contractors Association had begun to indicate that EPDM seams were failing and leading to roof leaks — not the sheets, the seams. Those seams were being made on the roof with liquid-applied polymer-based adhesives. What followed was a decade of public research into why, and then a change in practice.
| Study | What was examined | What it found | What it cannot tell you |
|---|---|---|---|
| Fort Belvoir, 1988 · NISTIR 88-3893 | Seams cut from an Army building whose EPDM laps had partially delaminated, tested for peel strength and surface condition against laboratory-made controls using the same rubber and adhesive. | Delamination within about a year of formation. Field peel strengths well below laboratory controls. Talc-like release agent on the bonding face, invisible to the eye. Small voids in the adhesive layer representing areas of no bond. In the lab: about 40 per cent strength loss with no cleaning, and more than 20 per cent loss curing at 158°F rather than 73°F. | How common this was. It is one building, described by its authors as a limited study. It establishes a mechanism, not a failure rate. |
| 48 field seams, 1991 · NIST symposium paper | Forty-eight adhesive-bonded seam samples taken from EPDM roofs, examined for adhesive identity, adhesive thickness and peel strength. | The majority of the seams performing unsatisfactorily had been bonded with neoprene-based adhesives and had been exposed 45 months or more. The butyl-based samples were rated satisfactory — but they were younger, and their peel strengths were only marginally better, so the paper does not separate chemistry from age. Most adhesive layers were thin, with no correlation found between thickness and performance. Microscopy again found release-agent residue on the majority of surfaces analysed. The paper concluded that a field method to judge surface cleanness was needed. | Which products are in front of you now. Both the adhesive chemistry and the dominant seam method have changed since. Only the abstract was publicly readable. |
| Fort Benning, 1991 · NISTIR 4525 | A 1980 EPDM roof inspected in January 1989, with existing repair patches sampled and two new patches made on the spot — one on a cleaned surface, one on an uncleaned one. | Butyl-tape patches were hard to peel by hand and tested about 6 lbf/in or better. Neoprene-adhesive patches peeled by hand, with friable, discoloured adhesive. Patches made from a grey caulk-like material no manufacturer recommends could not even be sampled. Loose patches had let water into wet insulation. | That cleaning does not matter. The new cleaned and uncleaned patches showed no statistically significant difference here — five replicates, one roof, one adhesive — and both contained visible areas of little or no bond. |
| EPDM Seam Consortium, 1997–98 · NIST BSS 176 and 177 | Laboratory creep-rupture testing of tape-bonded seams, funded by an industry–government consortium of adhesive and membrane manufacturers with NRCA and the Roof Consultants Institute, supported by the Army. Designed variables included tape type and thickness, surface condition, primer, application temperature, applied pressure and dwell time. | The move to tape was treated as a hypothesis and tested as one, against liquid-adhesive controls, across temperature extremes and industry protocols including cold-temperature preparation. | Field service life. This is accelerated laboratory loading, and only the abstracts were readable; the scanned reports carry no text layer. |
| Symposium summary, 1998 · J. Res. NIST 103(2) | Rossiter’s published summary of the Fourth International Symposium on Roofing Technology, including the consortium papers and a separate paper presenting one major manufacturer’s own warranty-repair-cost database, covering EPDM and polymer-modified bitumen installed from 1982 onward. | The main conclusion recorded: tape samples had mean times-to-failure that were, in most cases, comparable to or greater than those of the liquid-adhesive samples. The warranty database showed performance improvements attributed to changes including EPDM seam tapes in lieu of liquid adhesives. | That tape is unconditionally better. The finding is comparable-to-or-greater under test, and the warranty data is a manufacturer’s own. |
Read this table one item at a time
Fort Belvoir, 1988 · NISTIR 88-3893
- What was examined
- Seams cut from an Army building whose EPDM laps had partially delaminated, tested for peel strength and surface condition against laboratory-made controls using the same rubber and adhesive.
- What it found
- Delamination within about a year of formation. Field peel strengths well below laboratory controls. Talc-like release agent on the bonding face, invisible to the eye. Small voids in the adhesive layer representing areas of no bond. In the lab: about 40 per cent strength loss with no cleaning, and more than 20 per cent loss curing at 158°F rather than 73°F.
- What it cannot tell you
- How common this was. It is one building, described by its authors as a limited study. It establishes a mechanism, not a failure rate.
48 field seams, 1991 · NIST symposium paper
- What was examined
- Forty-eight adhesive-bonded seam samples taken from EPDM roofs, examined for adhesive identity, adhesive thickness and peel strength.
- What it found
- The majority of the seams performing unsatisfactorily had been bonded with neoprene-based adhesives and had been exposed 45 months or more. The butyl-based samples were rated satisfactory — but they were younger, and their peel strengths were only marginally better, so the paper does not separate chemistry from age. Most adhesive layers were thin, with no correlation found between thickness and performance. Microscopy again found release-agent residue on the majority of surfaces analysed. The paper concluded that a field method to judge surface cleanness was needed.
- What it cannot tell you
- Which products are in front of you now. Both the adhesive chemistry and the dominant seam method have changed since. Only the abstract was publicly readable.
Fort Benning, 1991 · NISTIR 4525
- What was examined
- A 1980 EPDM roof inspected in January 1989, with existing repair patches sampled and two new patches made on the spot — one on a cleaned surface, one on an uncleaned one.
- What it found
- Butyl-tape patches were hard to peel by hand and tested about 6 lbf/in or better. Neoprene-adhesive patches peeled by hand, with friable, discoloured adhesive. Patches made from a grey caulk-like material no manufacturer recommends could not even be sampled. Loose patches had let water into wet insulation.
- What it cannot tell you
- That cleaning does not matter. The new cleaned and uncleaned patches showed no statistically significant difference here — five replicates, one roof, one adhesive — and both contained visible areas of little or no bond.
EPDM Seam Consortium, 1997–98 · NIST BSS 176 and 177
- What was examined
- Laboratory creep-rupture testing of tape-bonded seams, funded by an industry–government consortium of adhesive and membrane manufacturers with NRCA and the Roof Consultants Institute, supported by the Army. Designed variables included tape type and thickness, surface condition, primer, application temperature, applied pressure and dwell time.
- What it found
- The move to tape was treated as a hypothesis and tested as one, against liquid-adhesive controls, across temperature extremes and industry protocols including cold-temperature preparation.
- What it cannot tell you
- Field service life. This is accelerated laboratory loading, and only the abstracts were readable; the scanned reports carry no text layer.
Symposium summary, 1998 · J. Res. NIST 103(2)
- What was examined
- Rossiter’s published summary of the Fourth International Symposium on Roofing Technology, including the consortium papers and a separate paper presenting one major manufacturer’s own warranty-repair-cost database, covering EPDM and polymer-modified bitumen installed from 1982 onward.
- What it found
- The main conclusion recorded: tape samples had mean times-to-failure that were, in most cases, comparable to or greater than those of the liquid-adhesive samples. The warranty database showed performance improvements attributed to changes including EPDM seam tapes in lieu of liquid adhesives.
- What it cannot tell you
- That tape is unconditionally better. The finding is comparable-to-or-greater under test, and the warranty data is a manufacturer’s own.
Read this table as a chain of custody rather than a scoreboard. Each row narrows the question: seams fail, they fail at the bond face, the bond face is contaminated in a way nobody can see, cleaning is unverifiable in the field, and a factory-made tape removes several of the field variables at once.
What actually changed, and what it means for a proposal
The substitution of factory tape for on-roof liquid adhesive is the most consequential thing that has happened to this system. It did not make the rubber better. It moved a set of variables — adhesive thickness, coverage, open time, solvent flash-off — out of a windy roof and into a factory. What it did not move is the first two items in the drawing above: the sheet still arrives with release agent on it, and somebody still has to clean and prime the bond face by hand, in the weather, on schedule.
That is why the useful question to a bidder is rarely “do you use tape?” NIST’s own retrospective records that tape-bonded seams are now routinely acknowledged by the industry as performing quite satisfactorily; the method is the default, and confirming it tells an owner very little. What tells them something is which cleaner, which primer, whose procedure, and who checks. The research says the remaining risk lives exactly there.
The honest limit of the long track record
EPDM’s trade association states that formulations have remained relatively constant for about forty years, and it is an association that exists to sell the point, so take the chronology and leave the adjective. Even granting it entirely, the argument cuts less far than it seems. If the sheet has not changed but the seam method has, then forty years of field performance is forty years of performance from a joint that no longer exists in that form for the first half of the period. A 1985 EPDM roof performing badly is weak evidence against EPDM today. A 1985 EPDM roof performing well is weak evidence for it.
The defensible version of the claim is narrower and still worth something: this is a cured rubber with a long, publicly investigated history, whose failure mode is known, located, and repairable. That is more than most building products can say.
Black or white is an energy and moisture decision, not a durability oneSection link
EPDM is the only common single-ply sold in both a deliberately dark and a deliberately reflective version of the same chemistry. That makes it the clearest place to look at what roof colour actually buys — and the arithmetic is less settled than either side of the argument admits.
Start with what is not in dispute. The Environmental Protection Agency’s heat-island compendium reports standard black asphalt roof surfaces reaching 165 to 185°F at midday in summer, while surfaces with both high solar reflectance and high thermal emittance peak at only 110 to 115°F. Those figures are for asphalt rather than for rubber, so treat them as the right order of magnitude rather than a measurement of an EPDM field. The direction is not in question: a black membrane runs far hotter than a white one, and everything underneath it feels that.
| The claim | What the evidence says | What an owner should do with it |
|---|---|---|
| It saves heating energy in cold climates | Directionally supported, magnitude contested inside the Department of Energy’s own laboratories. EPA states the heating penalty of a reflective roof is real but typically offset by summer savings, partly because winter days are short, the sun is low, northern winters are cloudier, and snow cover reduces the reflectivity difference anyway. But when LBNL ran DOE’s Roof Savings Calculator against its own earlier work, cooling savings agreed to within 15 per cent while heating penalties came out 6 to 12 times larger — up to 60 per cent of cooling savings against LBNL’s 5 per cent — flipping the sign in Chicago, New York, Philadelphia and Baltimore. | Do not settle this from a brochure or from this page. In a heating-dominated climate the honest position is that two credible federal tools disagree about whether reflectivity costs or saves, which is precisely the case for a building-specific energy model rather than a rule of thumb. |
| It runs hot, so the assembly dries itself | Supported, and this is the strongest technical argument for black. Department of Energy guidance states that dark membranes get very hot and the heat drives moisture back down into the building, and that mechanically attached dark membrane roofs in warmer climate zones avoided moisture problems for that reason. The same guidance says that with light-coloured membranes many roofs no longer get hot enough to do it, and that failures once limited to cold climates now happen in mixed ones. | Treat a colour change on an existing assembly as a moisture-design change. If the roof has been drying downward through a hot black membrane, the reroof needs an air and vapour strategy that does not rely on that heat any more. |
| It sheds snow faster | Not supported here. No source was found that establishes a snow-shedding difference at building scale, in either direction. What EPA does say is adjacent and cuts the other way: snow cover on a roof reduces the reflectivity difference between a cool and a non-cool roof, because both are white while the snow sits there. | Discount the argument in a bid conversation unless a bidder can produce evidence. And note that on a low-slope commercial roof, snow leaving faster is not automatically a benefit — where it goes, and what the drains do with it, is the design question. |
Read this table one item at a time
It saves heating energy in cold climates
- What the evidence says
- Directionally supported, magnitude contested inside the Department of Energy’s own laboratories. EPA states the heating penalty of a reflective roof is real but typically offset by summer savings, partly because winter days are short, the sun is low, northern winters are cloudier, and snow cover reduces the reflectivity difference anyway. But when LBNL ran DOE’s Roof Savings Calculator against its own earlier work, cooling savings agreed to within 15 per cent while heating penalties came out 6 to 12 times larger — up to 60 per cent of cooling savings against LBNL’s 5 per cent — flipping the sign in Chicago, New York, Philadelphia and Baltimore.
- What an owner should do with it
- Do not settle this from a brochure or from this page. In a heating-dominated climate the honest position is that two credible federal tools disagree about whether reflectivity costs or saves, which is precisely the case for a building-specific energy model rather than a rule of thumb.
It runs hot, so the assembly dries itself
- What the evidence says
- Supported, and this is the strongest technical argument for black. Department of Energy guidance states that dark membranes get very hot and the heat drives moisture back down into the building, and that mechanically attached dark membrane roofs in warmer climate zones avoided moisture problems for that reason. The same guidance says that with light-coloured membranes many roofs no longer get hot enough to do it, and that failures once limited to cold climates now happen in mixed ones.
- What an owner should do with it
- Treat a colour change on an existing assembly as a moisture-design change. If the roof has been drying downward through a hot black membrane, the reroof needs an air and vapour strategy that does not rely on that heat any more.
It sheds snow faster
- What the evidence says
- Not supported here. No source was found that establishes a snow-shedding difference at building scale, in either direction. What EPA does say is adjacent and cuts the other way: snow cover on a roof reduces the reflectivity difference between a cool and a non-cool roof, because both are white while the snow sits there.
- What an owner should do with it
- Discount the argument in a bid conversation unless a bidder can produce evidence. And note that on a low-slope commercial roof, snow leaving faster is not automatically a benefit — where it goes, and what the drains do with it, is the design question.
Reflectance also interacts with the adopted energy code, which in some climate zones credits or requires it. That is a jurisdiction question and it may remove the choice entirely.
The part of this that is not about energy at all
A hot black membrane and a cool white one place different demands on everything they touch. Higher membrane temperature means more thermal movement at terminations, hotter adhesives and sealants, and a hotter deck. A cooler membrane means the opposite, plus a surface that shows dirt, holds moisture longer after rain, and can support biological growth in humid climates. Neither of those lists is a verdict. They are inputs to a decision that also includes the energy code, the energy model, and the moisture behaviour of the assembly underneath.
The one framing to reject outright is that white EPDM is a “better” membrane than black EPDM. Both are the same cured rubber with the same seam requirements and the same vulnerabilities. What differs is how much of the sun’s energy the roof keeps, and what the assembly under it was designed to do with that energy.
Ballasted, mechanically attached, or fully adheredSection link
Attachment is a wind design produced for a specific building. It is also the choice that decides how a leak gets found, what the structure carries every day, and whether the roof can be inspected at all.
| Method | What holds the sheet | What it decides besides wind | Where it is the wrong answer |
|---|---|---|---|
| Ballasted | Weight. Round smooth stone, or pavers, laid loose over the membrane. Dated Canadian trade guidance puts conventional loose-laid gravel ballast at 50 to 65 kg/m² — 10 to 13 psf — varying by roof zone, using ASTM D448 gradation #2 or #4 stone. | Permanent dead load on the structure, every day of the year. Total loss of visual access to the membrane. Leak location becomes an excavation. Note also that a loose-laid membrane over an air-permeable deck is exposed to internal building pressure as well as suction, which increases the ballast required. | Any building whose structure has not been evaluated for the load. High-wind exposures where scour is a concern. Any portfolio where the owner needs to inspect roofs rather than react to them — which is most portfolios. |
| Mechanically attached | Fasteners and plates through the membrane into the deck, in rows, with the sheet spanning between fastener lines. The pattern and density are the design, and they are denser at perimeters and corners than in the field. | A continuous void between membrane and substrate, so water entering at one point can travel a long way before it appears inside. Also flutter: Department of Energy guidance attributes it to negative pressure above the membrane combined with positive pressure beneath it from interior air leaking into the assembly — which makes it an air-barrier problem as much as a fastener problem. | Over a deck that will not reliably hold fasteners. Where the wind design points to adhering instead — DOE guidance recommends fully adhered membranes in high-wind zones. And on any operating building where finding a leak quickly is worth more than the installed-cost saving. |
| Fully adhered | Adhesive across the whole area, so uplift is distributed rather than concentrated at fastener rows. It is what DOE guidance recommends in high-wind zones. | No plenum, so a leak surfaces closer to where it entered and costs far less to find. More material and more labour. Complete dependence on the substrate being sound, dry and compatible, and on the adhesive’s product-specific temperature window being met on the day. | Over a substrate that is not sound and dry. Where the schedule cannot move to meet the adhesive’s temperature limits. Over an existing bituminous surface without confirming, with the membrane manufacturer, whether a separator is required — at least one major manufacturer warns against direct contact between EPDM and bituminous systems. |
Read this table one item at a time
Ballasted
- What holds the sheet
- Weight. Round smooth stone, or pavers, laid loose over the membrane. Dated Canadian trade guidance puts conventional loose-laid gravel ballast at 50 to 65 kg/m² — 10 to 13 psf — varying by roof zone, using ASTM D448 gradation #2 or #4 stone.
- What it decides besides wind
- Permanent dead load on the structure, every day of the year. Total loss of visual access to the membrane. Leak location becomes an excavation. Note also that a loose-laid membrane over an air-permeable deck is exposed to internal building pressure as well as suction, which increases the ballast required.
- Where it is the wrong answer
- Any building whose structure has not been evaluated for the load. High-wind exposures where scour is a concern. Any portfolio where the owner needs to inspect roofs rather than react to them — which is most portfolios.
Mechanically attached
- What holds the sheet
- Fasteners and plates through the membrane into the deck, in rows, with the sheet spanning between fastener lines. The pattern and density are the design, and they are denser at perimeters and corners than in the field.
- What it decides besides wind
- A continuous void between membrane and substrate, so water entering at one point can travel a long way before it appears inside. Also flutter: Department of Energy guidance attributes it to negative pressure above the membrane combined with positive pressure beneath it from interior air leaking into the assembly — which makes it an air-barrier problem as much as a fastener problem.
- Where it is the wrong answer
- Over a deck that will not reliably hold fasteners. Where the wind design points to adhering instead — DOE guidance recommends fully adhered membranes in high-wind zones. And on any operating building where finding a leak quickly is worth more than the installed-cost saving.
Fully adhered
- What holds the sheet
- Adhesive across the whole area, so uplift is distributed rather than concentrated at fastener rows. It is what DOE guidance recommends in high-wind zones.
- What it decides besides wind
- No plenum, so a leak surfaces closer to where it entered and costs far less to find. More material and more labour. Complete dependence on the substrate being sound, dry and compatible, and on the adhesive’s product-specific temperature window being met on the day.
- Where it is the wrong answer
- Over a substrate that is not sound and dry. Where the schedule cannot move to meet the adhesive’s temperature limits. Over an existing bituminous surface without confirming, with the membrane manufacturer, whether a separator is required — at least one major manufacturer warns against direct contact between EPDM and bituminous systems.
The ballast figures above come from a 1993 Canadian trade bulletin that references standards since superseded. They are given as an order of magnitude and a zoning principle only. The current edition of ANSI/SPRI RP-4, the adopted wind-load standard, and the authority having jurisdiction govern any actual design.
A worked example: what ballast weighs
Take a 40,000 square foot warehouse roof, which is an ordinary size for a single-tenant distribution building. At the 10 to 13 pounds per square foot the Canadian bulletin describes for conventional loose-laid gravel ballast:
40,000 sq ft × 10 lb/sq ft = 400,000 lb = 200 tons
40,000 sq ft × 13 lb/sq ft = 520,000 lb = 260 tons
Between 200 and 260 tons of stone, sitting on the building permanently. That is dead load: it is there in July and it is there under whatever snow the structure is also required to carry. It is also the reason ballast is not a decision a roofing contractor makes on a reroof of an existing building. The Canadian guidance says so directly — the specifier must consider the implications of the weight on the structural capacity of the building and seek professional advice from a structural engineer if in doubt.
The second half of the example is the part that shows up in operating budgets rather than in structural drawings. To find a leak under ballast, someone moves stone. To inspect a seam under ballast, someone moves stone. Over twenty years that cost is paid repeatedly, usually at short notice, usually in bad weather, and it is almost never in the comparison that made ballast look cheap.
Puncture, traffic, and hailSection link
Away from the seams, the thing that ends an EPDM field sheet is almost always something that came through it from above — and most of those things arrived on a service technician’s route.
EPDM’s product standard sets both dynamic and static puncture resistance requirements, which is a real property of the sheet and a reason thickness is worth paying for on a trafficked roof. But the rating an owner can actually ask for lives at the assembly level. FM’s examination standard offers a Dynamic Puncture Resistance Rating of Roof Covers, tested to ASTM D5635, in steps from 5 to 50 joules. The important detail is that it is an optional rating. An assembly that does not carry one has not failed the test; it has usually not been submitted for it. Absence of a puncture rating tells you nothing, which is exactly why asking the question is worth something.
The same standard tests foot traffic by loading a three-inch square steel plate to 200 pounds, five times, and requires that the roof cover not tear or crack through to the substrate. Five loadings is a qualification threshold. A rooftop unit serviced quarterly for twenty years is eighty visits, each with a technician, a toolbag, and often a replacement part dragged across the membrane. This is why walkway pads belong on the drawing and not in the change-order pile, and why the route they follow should be the one a technician will actually take from the hatch to the unit.
Hail is the third mechanism and the one most often oversold. FM rates assemblies Moderate, Severe or Very Severe Hail. The Very Severe test fires two-inch preformed ice balls at 152 to 160 feet per second, which NRCA’s technical staff reported as 53 to 58 foot-pounds of impact energy, and the VSH designation applies to a narrow region: Oklahoma, Kansas and some northern Texas counties. When it was introduced, very few of the assemblies in FM’s listing database carried it. The same NRCA article cautions against describing components or systems as hail-resistant or hail-proof at all, because laboratory impact testing gives a relative measure that does not correlate to what happens on a real roof.
For an owner, that resolves into a short list. Thickness helps. A hard cover board under the membrane usually helps more, and it is one of the first items cut when a bid is being trimmed to win. Walkway protection helps most of all, because the failure that actually happens most often is not weather — it is a dropped tool on an unprotected path.
Where the money goes, and why no price per square foot appears hereSection link
No transparent national dataset separates installed EPDM cost from installed TPO or PVC cost. Publishing a number here would mean inventing one, and this page will not.
On a commercial low-slope roof the membrane is a minority of the price. The line items that move a number are the ones under and around it: tear-off versus recover and the disposal tonnage that follows; insulation type and thickness against the adopted energy requirement; tapered insulation if the deck does not already drain; a cover board; attachment density, which is a wind-design output; deck repair, which is the classic mid-project discovery; and the count of curbs, drains, penetrations and edge details, which is not proportional to area at all.
Two proposals for the same building can differ by more on insulation and cover board than the entire membrane line is worth. This is the specific reason a proposal that leads with a membrane type and a mil thickness is inviting a comparison on the wrong number.
Where EPDM has a genuine economic argument, it is on the operating side rather than the capital side: a system a competent crew can repair without a welder, on a roof someone can walk and inspect, tends to cost less to own than one where every repair is a specialist call-out. That advantage is entirely conditional on the roof being inspected and on repairs being made with the manufacturer’s own products. It disappears the moment a maintenance contractor starts patching with whatever is on the truck.
For how this site handles cost figures generally, see the roof cost hub and the cost methodology. For comparing what came back from a bid list, see comparing roofing quotes, and for framing the whole low-slope decision, commercial roofing.
What changes this on a real buildingSection link
- Code and jurisdiction
There is no nationwide building code for site-built construction in the United States. States and local governments adopt and amend model codes on their own cycles, so the edition that governs an EPDM reroof — and whether a recover is permitted at all over what is already there — is a question about a jurisdiction, not about a membrane. Energy-code requirements for insulation and, in some zones, for roof reflectance are adopted the same way.
Nothing on this page is a code determination. The adopted edition, the local amendments, the effective date, and the interpretation all belong to the authority having jurisdiction, and they should be confirmed with that office before a scope is written.- Fire
Class A, B and C describe a tested roof assembly, not a sheet. FM Approvals is explicit about the reason: the performance of a roof assembly depends on every component from the deck and its securement up to the uppermost weathering layer, so the entire assembly has to be evaluated as a single entity. On a low-slope roof that means the deck, the vapour or air control layer, the insulation, the cover board and the securement are all inside the tested combination.
A membrane does not carry a fire classification by itself. Ask which specific tested assembly is being installed, and confirm that the insulation and cover board being delivered are the ones in that listing.- Wind
Uplift performance is a property of an assembly on a particular building at a particular site. Basic wind speed, exposure, height, geometry, enclosure and risk category all feed it, and the required attachment density is higher at perimeters and corners than in the field. The dated trade guidance cited below already illustrates the zoning principle — corner zones sized from building height but never less than about eight and a half feet, and a perimeter band of similar minimum width.
A membrane has no wind rating on its own. FM and UL listings apply to specific tested assemblies at specific attachment densities. The design for a specific building is engineering work by a qualified professional, checked against the adopted code and the AHJ, and no table on this page substitutes for it.- Hail and impact
FM’s hail damage ratings run Moderate, Severe and Very Severe. The Very Severe Hail test fires two-inch preformed ice balls at 152 to 160 feet per second, producing 53 to 58 foot-pounds of impact energy, and NRCA’s technical staff recorded that VSH-classified assemblies were a very small fraction of the assemblies listed in FM’s database when the classification was introduced. The VSH region itself is narrow: Oklahoma, Kansas and some northern Texas counties.
No membrane is hail proof, and laboratory impact testing is a relative measure that does not correlate directly to what a real hailstorm does to a real roof. Treat any marketing use of the word hail-resistant as a claim about a test, and ask which test, on which assembly.- Moisture and ventilation
This is the consideration most often missed on an EPDM colour change. Department of Energy guidance describes dark membranes running very hot and driving moisture back down out of the assembly, and notes that in warmer climate zones mechanically attached dark membrane roofs avoided moisture problems for exactly that reason. Its next sentence is the warning: with energy conservation and light-coloured membranes, many roofs no longer get hot enough to drive that moisture back down, and failures that had been limited to cold climates now happen in mixed climates.
Changing a black roof to a white one changes the drying behaviour of everything under it. If the assembly relied on downward drying, that is a vapour and air-control question for the reroof design — see insulation, air and vapour control — not an afterthought.- Chemical exposure
EPDM has a documented vulnerability to animal fats, vegetable and animal oils, and petroleum-based products such as gasoline and other fuels. In NRCA’s technical review of the subject, one major membrane manufacturer — Johns Manville — is singled out for making special mention of avoiding direct contact with other roofing products such as bituminous systems, which matters on a recover over an existing built-up or modified-bitumen roof. That same review puts the responsibility on the roof system designer to establish what the building actually emits before a membrane is selected.
Chemical compatibility is product-specific. Some manufacturers publish resistance charts and some do not release them at all, so the only reliable answer is a written statement from the manufacturer about the substances on this roof.- Access and site conditions
Rooftop service traffic is a design input, not an operational detail. FM’s assembly examination includes a foot-traffic test in which a three-inch square steel plate is loaded to 200 pounds five times, and the assembly fails if the roof cover tears or cracks through to the substrate. A roof with mechanical units on it will see that load thousands of times over its life, along a path nobody drew.
Walkway pads are specified from where technicians will actually walk, which is the shortest line between the hatch and the equipment — not from where the drawing says the route is.- Maintenance
The 1988 NIST inspection of the Fort Belvoir roof ended with a recommendation that looks almost too obvious to state: thorough periodic inspection at six-month intervals, because acceptable roofing practice already emphasises it. The reason it was emphasised is worth carrying: the repairs on that roof were themselves the thing most likely to fail next, and nobody would know without looking.
What is covered, and what can actually be repairedSection link
Repairability is the strongest genuine claim EPDM has. It is also the claim most often overstated.
- Material-only versus system warranties
These are materially different documents. A material warranty is a promise about a sheet by the company that made it. A system or no-dollar-limit warranty covers a named assembly, usually requires a certified installer, and typically carries inspection and maintenance conditions that the owner has to perform. Which one a proposal includes changes what happens on the day of a leak more than any property of the membrane does. Read how a roofing warranty is actually read before comparing two of them on their headline year count.
- What the year number is not
A twenty-year warranty is a contract term. It is not a service life, not a prediction, and not a statement that the roof will need nothing for twenty years. Service life on this system is decided by the things in the diagram above and by whether anyone inspects the roof.
- Conditions that commonly bite
Documented inspections, prompt leak reporting, restrictions on other trades working on the roof, and rules about what may be attached to it. On EPDM specifically, adding equipment or a solar array after the fact usually involves someone cutting into the membrane, and who does that work is frequently the difference between a covered repair and an excluded one.
Repairability
A competent crew repairs EPDM with cleaner, primer, tape or adhesive, and a patch of the same rubber. No welder, no power, no generator on the roof. Against a thermoplastic system, where the repair depends on a working hot-air welder and someone who can set it correctly, that is a real operational advantage for a portfolio with a leak-response programme.
The federal record also shows the honest limit of it. When NIST inspected a 1980 EPDM roof at Fort Benning in January 1989, it found patches made with three different things. The butyl-tape patches were well bonded and hard to peel by hand, and tested at about 6 pounds per inch of peel strength or better. The patches made with neoprene-based adhesive peeled off by hand with little force, and the adhesive underneath had gone friable and mustard-brown. Some patches had been made with a grey caulk-like material that was not a product any manufacturer recommended for the purpose — those were so poor that specimens fit for laboratory testing could not be obtained from them. Loose patches had let water in, and the insulation below them was wet.
The lesson is not that EPDM is hard to repair. It is that an easy repair is easy to do badly, and that a bad patch is invisible from a photograph and from the ground. If a building has a maintenance contractor patching its roof, the question worth asking is which specific products they use and whether those products are the ones the membrane manufacturer names.
A warranty is a contract between a reader and whoever wrote it. What it covers, what voids it, whether it transfers, and how it is enforced are set by that document and by the law where the reader lives. Read the actual warranty for the product and the installer in front of you — not a summary of one, including this one.
Questions for a commercial roofing bidderSection link
Every one of these has a specific right answer. A bidder who cannot give one is bidding a sheet, not a roof.
What is the seam method, and which cleaner and primer are specified — by product name, from the membrane manufacturer?
This is the question the whole page is about. A good answer names the tape, the cleaner, the primer and the manufacturer’s own procedure. A vague answer means the most consequential variable on the roof is being left to whoever shows up.
What is your crew’s procedure for rolling seams, and how is it checked before the day’s work is closed out?
Pressure is a designed variable in the federal research and a human one on the roof. Ask how seams are probed, who probes them, and what gets written down. Probing is cheap; finding an unprobed seam three winters later is not.
What happens to seam work when the temperature drops or the surface is not dry, and who has authority to stop?
NIST’s consortium work treated application temperature and dwell time as designed variables, and its earlier laboratory work found bond strength falling by more than 20 per cent under an elevated cure temperature alone. Adhesives and tapes carry product-specific temperature windows. A crew under schedule pressure with no stop authority is the mechanism by which those windows get ignored.
Which specific tested assembly are we buying, and does the insulation and cover board on the delivery ticket match it?
Fire and uplift classifications belong to assemblies, not sheets. A substituted cover board is a substituted assembly, and the substitution is usually made after the bid is won.
What does the wind design require at the corners and perimeter, and who produced it?
The correct answer names a qualified designer and a document. An answer that is a rule of thumb, or a number remembered from another job, is a signal that the densest attachment on the roof is being guessed at.
Where are the walkway pads going, and how did you decide the route?
The right answer traces the line a technician will actually walk from the hatch to each unit. The wrong answer is that walkways were not included, which usually means the traffic damage has been deferred to the owner.
If this is a recover, what is under it, is it wet, and how do you know?
Trapped moisture under a new membrane is a warranty and a structural problem at once, and at least one major EPDM manufacturer warns against direct contact between its membrane and bituminous systems — so ask yours, in writing. The answer should reference a condition assessment with core cuts or a moisture survey, not an opinion formed from the parapet.
Require these in writing
- Membrane type, ASTM D4637 Type, thickness in mils, and colour, stated separately
- Seam method, and the cleaner, primer, tape or adhesive and lap sealant by product name
- The specific tested assembly designation being installed, with every component named
- Attachment method and the fastening or adhesive pattern for field, perimeter and corner zones
- Insulation type and thickness, cover board, and whether either may be substituted
- Termination detail at every parapet, curb, drain, scupper and penetration
- Walkway pad locations, shown on a drawing
- Who probes seams, when, and what documentation the owner receives at closeout
- The warranty document itself, not a summary of it — including its inspection and maintenance conditions
Misconceptions and failure modesSection link
Common misconceptions
Common belief
EPDM has the longest track record, so it is the safest choice.
What is actually true
The long record is real, and it is the strongest thing this material has. But the record is not uniform: the first two decades of it are largely a record of seams bonded with liquid adhesive, a method that has been superseded. Field evidence about 1980s EPDM is weak evidence about the EPDM being quoted now, in the same way that field evidence about early thermoplastics is weak evidence about the thermoplastics being quoted now. Track record is a question about a specific product and a specific method, not about three letters.
Common belief
Black roofs are better in cold climates.
What is actually true
The direction of that argument is defensible; the size of it is genuinely contested inside the Department of Energy’s own laboratories. See the colour section above for what the two federal simulation tools actually disagreed about, and by how much. What is not contested is the moisture consequence, which is a better reason to take the colour decision seriously than the energy arithmetic is.
Common belief
EPDM cannot be repaired once it is old.
What is actually true
It can, and repairing it is one of the reasons owners choose it. What the federal work found is narrower and more useful: bonding to an aged, weathered sheet depends on getting that surface genuinely clean, and cleanliness is hard to verify in the field. Where NIST tested new patches on an aged in-service membrane, both the cleaned and the uncleaned specimens contained visible shiny areas indicating regions of little or no bond.
Common belief
A ballasted roof is the cheap option.
What is actually true
It is often the cheapest to install and among the most expensive to own. Ballast is permanent dead load the structure carries every day. It hides the membrane, so a leak cannot be found without moving stone. And it makes routine inspection — the practice that actually determines service life on this system — expensive enough that it tends not to happen.
Common belief
A thicker membrane is a longer-lasting roof.
What is actually true
Thickness buys puncture and impact margin, which is worth having. It does not buy seam quality, attachment adequacy, drainage, or termination detail, and those are what the evidence on this page is about. A 90-mil sheet with a badly prepared lap is a 90-mil sheet with a leak.
How it actually fails
- Seam delamination at a liquid-adhesive lap
- The classic first-generation failure. NIST’s inspection of an Army building at Fort Belvoir, Virginia found seams that had partially delaminated within about a year of being formed, with field peel strengths far below what the same adhesive and rubber achieved in the laboratory. Two mechanisms were implicated: talc-like release agent still on the bonding face, and small voids in the adhesive layer that represented areas of little or no bond.What you can see: Open or lifting lap edges, a lap that can be started with a fingernail, staining that follows the line of a seam rather than pooling around a penetration.
- The patch is the leak
- Repairs made with the wrong product, or bonded to an aged surface that was never properly cleaned. At Fort Benning NIST found patches that peeled by hand, and patches made from a caulk-like material no manufacturer recommends for the purpose. Where those patches were loose, the insulation underneath was wet.What you can see: A history of repeat repairs in the same area, mismatched patch material, sealant applied over a patch edge as a second attempt, and a soft or spongy feel underfoot near old repairs.
- Chemical attack from the building’s own exhaust
- Fats and oils from a kitchen exhaust, or hydrocarbons from process equipment, settling on a membrane documented as vulnerable to both. This is not a defect; it is a selection error, and it is usually visible as a defined plume area downwind of a fan.What you can see: A discoloured, softened or swollen area of field membrane with a shape that matches a fan’s prevailing plume, often with the worst damage nearest the unit.
- Traffic damage on an unplanned route
- Concentrated loads from service technicians, tools and equipment on a path that was never protected. FM’s foot-traffic acceptance criterion is that the roof cover does not tear or crack through to the substrate under a 200-pound load on a three-inch plate; a rooftop unit serviced quarterly for twenty years applies that far more often than five times.What you can see: A visible worn line between the hatch and the equipment, crushed insulation underfoot, punctures clustered around units rather than distributed across the field.
- Moisture accumulation after a colour change
- An assembly that had been drying downward through a hot dark membrane stops doing so when a reflective membrane is installed over the same construction, and the vapour and air control strategy underneath was not revisited. Department of Energy guidance names this directly, and notes the failures moving from cold climates into mixed ones.What you can see: Rising energy use, condensation appearing at fasteners or deck flutes, and moisture found by a survey in an assembly with no identifiable leak path.
Sources and further readingSection link
Understanding Roofing
Scope and limitations
- It cannot tell you whether your structure can carry ballast, or any other dead load.
- That is a determination a licensed structural engineer makes for this building, from this building's drawings and condition.
- It publishes no price per square foot and no service life in years.
- Both vary more with detail count, attachment, traffic and maintenance than with the membrane, and no transparent national dataset separates EPDM from its alternatives.
- It is not a code determination.
- Adopted edition, local amendments and effective dates are set by the authority having jurisdiction, and every code-adjacent statement here has to be confirmed with that office.
- It cannot tell you whether a specific assembly carries an FM or UL listing.
- That lives in the listing agency's own database and in the manufacturer's assembly documentation, tied to named components.
- It cannot resolve the black-versus-white energy question for your building.
- The federal simulation tools disagree, and the answer depends on your climate, your insulation, your fuel prices, and your HVAC — which is what an energy model is for.
- It cannot tell you whether a black membrane will clear snow faster than a white one on your roof.
- No source was found that would support a claim in either direction at building scale.
- It is not a warranty interpretation.
- What a specific warranty covers, what voids it, and whether it transfers are set by that document and by the law where the building stands.
Factors Affecting the Service Life of Seams of EPDM Roof Membranes
Walter J. Rossiter, Jr. and Jonathan W. Martin, National Institute of Standards and Technology; Proceedings of the 10th International Conference on Durability of Building Materials and Components, Lyon, France / 17–20 April 2005
NIST's own retrospective account of why this research programme existed: that EPDM captured about one-third of the total low-slope roofing market by the mid-1980s; that also by the mid-1980s, surveys conducted by the U.S. National Roofing Contractors Association began to indicate that EPDM seams were failing and leading to roof leaks; that those seams were fabricated in the field using liquid-applied polymer-based adhesives; that the majority of the research was conducted under an industry–government consortium; and that EPDM seams fabricated with tape adhesives are today routinely acknowledged by industry as performing quite satisfactorily.
Only the abstract was read, from NIST's own bibliographic record. It is a retrospective overview written by the lead researcher, so its account of how well the recommendations worked is not an independent assessment. It gives no failure rates and no service life.
Report of Roof Inspection: Partial Delamination of Adhesive-Bonded Seams at an Army Facility (NISTIR 88-3893)
Walter J. Rossiter, Jr. and James F. Seiler, Jr., National Institute of Standards and Technology, for the U.S. Army Engineer District, Baltimore / November 1988
Seams at Fort Belvoir, Virginia partially delaminated within about a year of formation; field specimens had low T-peel strengths against laboratory-prepared controls; SEM found talc-like contamination not visible to the eye, and small voids in the adhesive layer; not cleaning the rubber reduced seam strength about 40 per cent, and curing at 158°F rather than 73°F reduced it more than 20 per cent; the report concluded a means for assuring the quality of cleaned rubber was needed, and recommended six-month inspection intervals.
One building, one adhesive and rubber system, a limited study by its own description. It establishes mechanisms, not failure rates for EPDM as a category.
Cleaning of Aged EPDM Rubber Roofing Membrane Material for Patching: Analytical Techniques for Surface Characterization (NISTIR 4525)
Walter J. Rossiter, Jr. and Tinh Nguyen, National Institute of Standards and Technology, for the U.S. Army Construction Engineering Research Laboratory / February 1991
Inspection on 25 January 1989 of an EPDM roof built in 1980 at Fort Benning, Georgia: original neoprene-adhesive seams had peel strengths slightly below 1 lbf/in while butyl-tape patches measured about 6 lbf/in or greater; neoprene patches peeled by hand and their adhesive had gone friable and mustard yellow-brown; patches made from a grey caulk-like material not recommended by any manufacturer could not yield testable specimens; loose patches had admitted water and the insulation below was wet; new patches made during the visit on cleaned and uncleaned aged membrane showed no statistically significant difference in that test, and both contained shiny areas indicating little or no bond.
An interim report on surface-analysis method development, with a small number of specimens from a single roof. The cleaned-versus-uncleaned result is from five replicates and should not be read as evidence that cleaning does not matter.
Characteristics of Adhesive-Bonded Seams Sampled From EPDM Roof Membranes
W. J. Rossiter, Jr., J. F. Seiler, Jr., W. P. Spencer, J. A. Lechner and P. E. Stutzman, National Institute of Standards and Technology; presented at the 3rd International Symposium on Roofing Technology / 17–19 April 1991
Forty-eight seam samples were examined; the majority of the unsatisfactorily performing seams had been bonded with neoprene-based adhesives and had been exposed 45 months or more; the butyl-based samples were rated satisfactory but were younger and only marginally stronger in peel; the majority of samples had relatively thin adhesive layers (less than 0.13 mm) with no correlation between thickness and performance; microscopy showed evidence of release agent on the majority of surfaces analysed; and the paper concluded that a field method to judge rubber surface cleanness before adhesive application is needed.
Only the abstract was read, from NIST's own bibliographic record. The full paper is in the symposium proceedings and was not available.
Consortium to evaluate performance of tape-bonded seams for sealing EPDM roofing
National Institute of Standards and Technology / 11 April 1995
The industry–government EPDM Seam Consortium and its members: ADCO Products, Ashland Chemical, Carlisle SynTec Systems, Firestone Building Products, Genflex Roofing Systems, the National Roofing Contractors Association and the Roof Consultants Institute, with the U.S. Army Construction Engineering Research Laboratories as a supporting sponsor. Also that concern about volatile organic compounds in solvent-based adhesives was part of what drove industry interest in tapes.
Performance of Tape-Bonded Seams of EPDM Membranes: Effect of Material and Application Factors on Peel Creep-Rupture Response (NIST BSS 176)
W. J. Rossiter, Jr., M. G. Vangel, K. M. Kraft and J. J. Filliben, National Institute of Standards and Technology / May 1997
The variables the consortium treated as designed factors: two tape types at 0.9 mm and 0.6 mm thickness, surface condition, primer use, application temperature of 5°C or 60°C, applied pressure of 0.2 or 2 MPa, and duration at temperature from 24 to 960 hours, benchmarked against liquid-adhesive sample sets from an earlier phase.
Only the abstract was read; the scanned full report has no usable text layer. Laboratory creep-rupture testing, not field performance.
Performance of Tape-Bonded Seams of EPDM Membranes: Factors Affecting the Creep-Rupture Response of Tape-Bonded and Liquid-Adhesive-Bonded Seams (NIST BSS 177)
W. J. Rossiter, Jr., M. G. Vangel and K. M. Kraft, National Institute of Standards and Technology / July 1998
Phase III of the consortium work: five tasks covering elevated test temperature, heat exposure before loading, Rubber Manufacturers Association and SPRI industry protocols, cold-temperature preparation, and shear testing, comparing two commercial tape systems and one liquid adhesive on cleaned EPDM.
Only the abstract was read; the scanned full report has no usable text layer.
Conference Report: Fourth International Symposium on Roofing Technology (Journal of Research of NIST, Volume 103, Number 2)
Walter J. Rossiter, Jr., Building and Fire Research Laboratory, National Institute of Standards and Technology / March–April 1998
That adhesive tapes had begun to supplant contact-type liquid adhesives in EPDM seams, and that a main conclusion of the consortium creep-rupture study was that tape samples had mean times-to-failure that were, in most cases, comparable to or greater than those of the liquid-adhesive samples. Also that a manufacturer's warranty-repair-cost database covering EPDM and polymer-modified bitumen installed from 1982 onward showed performance improvements attributed to changes including the use of EPDM seam tapes in lieu of liquid adhesives.
A summary of symposium papers rather than the papers themselves. The warranty-database finding is a manufacturer's own data as reported at the symposium.
ASTM D4637/D4637M-15(2021)e1 — Standard Specification for EPDM Sheet Used in Single-Ply Roof Membrane
ASTM International / 2015, reapproved 2021 with editorial change
Type I non-reinforced, Type II internally reinforced with fabric or scrim, Type III fabric-backed; the properties the specification sets, including sheet and coating thickness, breaking and tensile strength, dynamic and static puncture resistance, tear resistance, brittleness point, ozone resistance, heat aging, factory seam strength and weather resistance; and that the specification establishes minimum sheet quality while excluding in-place roof system considerations such as fire resistance and uplift performance.
The abstract and scope were read from ASTM's public product record. The full standard is paywalled and its numeric property limits were not read.
Examination Standard for Class Number 4470: Single-Ply, Polymer-Modified Bitumen Sheet, Built-Up Roof (BUR) and Liquid Applied Roof Assemblies for Use in Class 1 and Noncombustible Roof Deck Construction
FM Approvals LLC / April 2022
That the performance of a roof assembly depends on all components from the deck and its securement to the uppermost weathering layer, so it is necessary to evaluate the entire roof assembly as a single entity; that certification covers named assembly combinations with every component identified by manufacturer, trade name and installation method; the rating families available, including ASTM E108 Class A, B or C with maximum roof slope, wind uplift classes, hail damage ratings of Moderate, Severe or Very Severe, and an optional Dynamic Puncture Resistance Rating of 5 to 50 joules tested to ASTM D5635; and the foot-traffic test, in which a three-inch square steel plate is loaded to 200 pounds five times and the cover must not tear or crack through to the substrate.
A certification standard, not a code and not a design method. An approval applies to the tested assembly, not to any component in isolation.
Understanding FM VSH
Mark S. Graham, Professional Roofing (National Roofing Contractors Association) / December 2017
The Very Severe Hail test method — two-inch preformed ice balls at 152 to 160 feet per second, producing 53 to 58 foot-pounds of impact energy; the VSH region of Oklahoma, Kansas and some northern Texas counties; how few assemblies carried the classification when it was introduced; and the caution against representing components or systems as hail-resistant or hail-proof, since laboratory testing gives a relative measure that does not correlate to actual rooftop conditions.
Written in 2017. The number of VSH-classified assemblies has almost certainly changed; treat the count as historical context and check the listing database for a current answer.
Chemical considerations
Professional Roofing (National Roofing Contractors Association) / November 2017
That EPDM membranes have vulnerability to animal fats, vegetable and animal oils, and petroleum-based products such as gasoline and other fuels; that one named manufacturer, Johns Manville, makes special mention of avoiding direct contact with other roofing products such as bituminous systems — the article attributes that caution to that manufacturer, not to the category; that the roof system designer bears responsibility for establishing what the building emits; and that manufacturers differ in whether they publish chemical-compatibility information at all.
Low-Slope (Flat) Roofs
Building America Solution Center, Pacific Northwest National Laboratory for the U.S. Department of Energy
That fully adhered membranes are recommended in high-wind zones; that membrane flutter is caused by negative pressure above the membrane combined with positive pressure beneath it from interior air leaking into the assembly; that dark membranes get very hot and the heat drives moisture back down into the building, which is how mechanically attached dark membrane roofs in warmer climate zones avoided moisture problems; and that with energy conservation and light-coloured membranes many roofs no longer get hot enough to drive that moisture back down, so failures once limited to cold climates now happen in mixed climates.
Written primarily for residential and small commercial construction, and it does not discuss ballasted systems at all.
Cool Roofs (Reducing Urban Heat Islands: Compendium of Strategies, Chapter 4)
U.S. Environmental Protection Agency / 2008
That standard black asphalt roofs can reach 165 to 185°F at midday during summer while cool roofs with both high reflectance and high emittance reach peak temperatures of only 110 to 115°F; that conventional roofs can be 55 to 85°F hotter than the air while cool roofs stay within 10 to 20°F of it; that EPDM is listed among common cool single-ply materials, described as a synthetic rubber with seams that must be glued or taped; and that the wintertime heating penalty is generally outweighed because winter days are shorter and the sun lower, with more winter cloud in mid-Atlantic and northern states, and because snow cover on roofs in those climates reduces the reflectivity difference between cool and non-cool roofs.
Published in 2008, and its temperature figures are for black asphalt roof surfaces rather than measurements of EPDM. Its energy and cost tables predate current energy prices and current energy codes.
Using Cool Roofs to Reduce Heat Islands
U.S. Environmental Protection Agency
The roles of solar reflectance and thermal emittance, and the statement that because cool roofs reflect sunlight and reduce solar heat gain they may increase energy use during winter months in cold climates, a heating penalty that is typically offset by summer cooling savings.
Analysis of DOE's Roof Savings Calculator with Comparison to other Simulation Engines (ORNL/TM-2013/501)
Joshua New, William A. Miller, Yu Huang, Ronnen Levinson, Jibonananda Sanyal and Kenneth Childs, Oak Ridge National Laboratory / October 2013, issued January 2014
That when Lawrence Berkeley National Laboratory's Heat Island Group used the Roof Savings Calculator, cooling savings agreed with earlier LBNL studies to within 15 per cent but heating penalties came out 6 to 12 times larger; that the RSC showed average heating penalties up to 60 per cent of cooling savings where the LBNL study showed 5 per cent; and that the RSC therefore showed cool roofs to be detrimental in colder locations such as Chicago, New York, Philadelphia and Baltimore, where LBNL's earlier study had found them beneficial in all 14 US climates studied.
A comparison between simulation engines, not a field study, and much of the disagreement traces to how attic radiant heat exchange is modelled. It says nothing about EPDM specifically. Treat it as evidence that the magnitude of the winter penalty is contested, not as a number to put in a business case.
Design of Loose-Laid Gravel Stone Ballasted Roofs (Volume 40)
Canadian Roofing Contractors Association / November 1993
That ballast in conventionally loose-laid gravel-stone single-ply assemblies runs from 50 to 65 kg/m² (10 to 13 psf) and varies by roof zone; that corner zones are sized at 40 per cent of building height but not less than 2.6 m (8.5 ft) and perimeters are a minimum 2.6 m wide; that ballast is assumed to be round smooth stone of ASTM D448 gradation #2 or #4; that an air-permeable deck exposes the loose-laid membrane to internal pressure as well as suction, increasing the ballast required; that mechanical ventilation and stack effect add internal pressure; that the recommended amount of ballast may leave the membrane visible, which is not a design flaw; and that the specifier must consider the weight against the building's structural capacity and seek a structural engineer if in doubt.
Dated 1993 and Canadian. It references ANSI RP-4 and ASCE 7-85/7-88 as they stood then; ANSI/SPRI RP-4 has been revised repeatedly since, and the adopted wind-load standard has changed. Use it for the order of magnitude and the zoning principle only. It is not a current design basis anywhere.
29 CFR 1910.28 — Duty to have fall protection and falling object protection
U.S. Occupational Safety and Health Administration
The general-industry duty to protect each employee on a walking-working surface with an unprotected side or edge four feet or more above a lower level using guardrail, safety net or personal fall-protection systems, and the low-slope roof provisions that set requirements by distance from the roof edge, with the less protective options available only where the work is both infrequent and temporary.
Federal general-industry text. State plans may impose more stringent requirements, and construction work is governed by a different subpart.
EPDM historical timeline
EPDM Roofing Association
The chronology only: that seam tapes came in during the 1980s to replace liquid adhesives, that white EPDM and 90-mil sheet are later additions to the category, and the association's own claim that EPDM formulations have remained relatively constant for about forty years.
A trade association that exists to promote this material. Its chronology is usable; its characterisations of performance are not independent, and it gives no date for the introduction of white EPDM.