GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Audio Play Icon Headphones Icon Plus Icon Minus Icon Check Icon Print Icon Picture icon Single Arrow Icon Double Arrow Icon Hamburger Icon TV Icon Close Icon Sorted Hamburger/Search Icon

Community and Q&A

Best/cheapest insulation method for flat roofs in new construction?

Scott West | Posted in Energy Efficiency and Durability on

I am an architect in Atlanta, GA (climate zone 3) mostly designing modern/contemporary homes that typically have flat roofs. There seems to be a lot of conflicting information and opinions out there as to the best way to insulate these types of roofs. Typical construction would be 12″ deep ceiling/roof I-joists or 2×4 open-web trusses with flat OSB decking above. The most common potential options (all unvented scenarios)are:

1. Cheaper route that many builders will generally favor: Build and additional layer of sloped OSB decking on ripped 2x furring strips for slope to roof drains and place R-30 fiberglass batts between rafters. Not particularly smart or green in any way and requires extensive caulk air-sealing but as with most things, the most-common way is most-affordable (up-front cost at least). Other potential downside is that if ceiling plane is not adequately air-sealed (for instance at recessed light fixtures) then warm air at underside of deck could cause condensation in colder months.

2. Place average 4″ of XPS rigid insulation on top of the decking (~R-20) and 4″ of closed-cell foam (R-24)doubling as air-sealer at underside of deck. Total R-value = ~44.

3. Place average 4″ of XPS rigid insulation on top of the decking (~R-20) and 2″ of closed-cell foam (R-12)doubling as air-sealer at underside of deck and then R-30 batts. Total R-value = ~62.

4. Forgo XPS insulation above deck and used sloped framing/decking as in option 1 but use 2″ of cc foam (R-12) add fill out remaining 10″ of cavity with dense-packed cellulose (~R-35). Total R-value = ~R-47

I know there’s a lot of variables here and the answer is often “it depends on priorities” but just looking to get some feedback on best balanced approach taking into account some of the following concerns:

– What scenario is going to provide best R-value for the money?

– Is using closed cell foam on underside of roof deck a real problem in the event of trying to track down and fix a potential roof leak? Is open-cell a better alternative on this count despite the lower R-value but with added benefit of also being cheaper?

– Will the weight of10″ of cellulose cause deflection of the drywall ceiling over time or it remains pretty self-supporting?

– Recessed can lights can be reduced but seem unavoidable in the ceiling plane. What factors to consider due to this?

– For considering cost trade-offs between the various options, I would assume the value of adding the extra layer of sloped framing/decking as in options 1 and 4 to average maybe +/-$1/sf of roof.

– On most projects, the builders seem to say that doing sloped XPS insulation is more expensive than having the framer build the slopes but not sure of the exact cost differential.

– Many of my clients would want to take into account the ‘green-ness’ of various products as a factor as well.

Any help in assessing these options and especially in comparing real-world costs would be greatly appreciated.


Scott West

GBA Prime

Join the leading community of building science experts

Become a GBA Prime member and get instant access to the latest developments in green building, research, and reports from the field.


  1. Jin Kazama | | #1

    Scott: i am no expert , but much interested in flat roofs/decks.

    From, your location doesn't seem to have cold temps much ??

    The required R value will guide the design of the roof much if you ask me.
    Do you have access to "green" SPUF ?
    Shooting foam from under the deck to seal and insulate ( in conjunction with upper insulation )
    would be something good,only if it is not done using shitty blowing agents.

    Also the sizes that you will be playing with would change the type of roofs much.

    I would tend to say that sloped EPS or ISO
    ( don't use XPS ... more expensive and less green ) is hard to beat on price
    as you are getting a slope and insulation in the same product.
    Also it is much faster to screw through 2 layers of foam than it is to fabricate a slopped lumber deck
    + second plywodd + sealing issues.

    Simplicity is a must here.

    Personally, if budget and size is good for,
    i'd go with steel web joists with steel deck on it
    Then peel stick membrane on top of the deck and onto the walls
    then screw in polyiso board + polyiso slopes
    lastly EPDM + ballast or fully adhered ..white if required to reflect sun

    But i seriously don't know the requirements for a flat roof in your town,
    any risks of condensation onder the deck ?

    BTW: cellulose is not self supporting on vertical, it will load the gypsum and needs to be calculated,
    unless the "bags" are fastended to the joists and the spacing is not great enough for the bulge to push down on the boards

    Pretty easy to find roof screws up to 8" in length at cheap price
    so 7XR 5.5 for long term iso gives R38 + the recover board could be iso also ...near R40

  2. GBA Editor
    Martin Holladay | | #2

    The most important piece of information you provided was this: "Recessed can lights ... seem unavoidable in the ceiling plane."

    It's too bad that you (as an architect) don't have the ability -- either the persuasive power or the judgment -- to insist that an insulated ceiling have zero recessed can lights.

    If the can lights are there, it clearly makes no sense whatsoever to depend on fluffy insulation (either fiberglass or cellulose) sitting on top of the ceiling drywall. All of those options should be taken off the table.

    That leaves either spray foam insulation on the underside of the roof sheathing, or rigid foam insulation on top of the roof sheathing.

    In case you missed it, here is an article that you probably want to read: Insulating Low-Slope Residential Roofs.

  3. Expert Member
    Dana Dorsett | | #3

    In Atlanta even an inch or two of exterior rigid polyiso or 3" of rigid high density rigid rock wool panels would be enough dew point control to allow you to blow the joists full of cellulose. Alternatively, 2" of closed cell foam on the underside of the roof deck would be sufficiently protective to fill the rest with blown fiberglass. See tables 3 & 4 in this document:

    I have yet to see batts that fit well with I-joists, and will have more installation problems than they're worth. Using blown fiber eliminates the performance robbing gaps and compressions that create thermal bypass paths. Cellulose is more protective of the structural wood since it buffers seasonal moisture. In a ceiling app it's worth dense-packing to at least 3lbs/cubic-foot density, which in some cases may make it more expensive as open cell foam, in which case o.c. foam may be a more attractive (if somewhat less green) solution. Get it quoted with 3lb cellulose, 1.8lb Optima/Spider/L77 noo-skool fiberglass, or half-pound foam- any of which would be a huge performance improvement over any batt solution.

    What Jin said about XPS vs. polyiso for the rigid foam- the blowing agents used for XPS have 200x the global warming hit of those used for polyiso, (and at R20 may have a higher global warming hit than the energy use it offsets) but rigid rock wool is probably greener overall, and more fireproof (and yes, more expensive.)

    So in order:

    Option 1 fails due to low- compatibity of batts to the application and lower overall performance

    Option 2 fails due to high GWP of both the exterior foam and the 4" of closed cell- this is guaranteed to be net-negative for the environment on a lifecycle basis.

    Option 3 fails for all of the reasons in 1 & 2

    Option 4 makes it, but you'd be better off with 2-4" of exterior iso (greenest) or 2" of 3lb spray polyurethane. + full cavity fill of cellulose. If you use the 3lb goods on the exterior it has higher GWP than polyiso (but will eventually break even) and the foam itself can be utilized as the finish roofing provided you give it a UV resistant coating, saving a bit on upfront cost and eventual roofing-replacement costs. (Mopping on new UV-resistance goop every 15-25 years is a lot easier & cheaper than re-roofing with membanes or composite asphalt solutions.)

  4. Scott West | | #4

    Thanks for all the comments!

    Jin - What is green SPUF? Is that an acronym for soy-based foam? FYI , steel framing and decking is not typically an option in the budget price-points of my projects. These are usually all wood construction homes. My brief research indicated that iso was a good bit more expensive than XPS but maybe that's not true? And I think EPs is too soft as we are often 'floating' rooftop decks over some parts of these roofs.

    Martin - I try to limit the number of recessed cans as much as possible and use many fewer than most architects I know but the reality is that people that want a modern home tend to also value having wall art and they want to light it appropriately and there's no other decent/affordable way to do that except with recessed can products. And most clients aren't willing to pay the significant extra cost for LED's. If say 3% of the ceiling plane insulation is compromised by can lights is the final effect really much more significant? Keep in mind that in most of the scenarios I proposed that the 2-4" of foam in underside of the decking would still be continuous as the can lights are usually only about 6" deep.

    Dana - Can rockwool panels be cut to a slope as is common with the foam boards for roof drainage? Never heard of this being done. I guess in the option 4 variation you outline, the spray polyurethane provides the required air sealing whereas going with any board product on the roof some kind of air-sealing has to happen below the roof deck which another extra cost if there's only cellulose below. Any idea of the ballpark cost on spray polyurethane for roofs as I have never used it before? Is it common enough and reasonably affordable?

    Dana, you have come closest but still hoping someone can chime in on what is likely best balance between real-world cost, sustainability, and energy savings. The reality with most clients, for better or worse, is that they can be much more easily convinced to invest in energy (and hence money) saving features before sustainability.

    Thanks again all!

  5. Expert Member
    Dana Dorsett | | #5

    If by "...cut to a slope as is common with the foam boards for roof drainage..." you mean manufactured with a taper for drainage, SFAIK no- they are only available as even-thicknesses, not tapered goods.

    In terms of real-world costs and sustainability, tapered EPS or polyiso for drainage, with the thin edge of the foam being no thinner than 1.5" if EPS, 1" if iso, and a full cavity fill of cellulose between the rafters is probably your best sustainable bang & performance buck at a level that still has some long-term rationale on cost. With 12" TJIs for rafters that would yield about an ~R50 "whole-assembly R", with a comparatively low environmental hit from the materials used. (The cellulose is mostly recycled materials, the EPS is blown with low GWP blowing agents, and allows you to use cellulose for the majority of the insulation without complicated venting schemes.)

    For the rationale on the foam thickness refer to TABLE R806.4 in the IRC:

    Note, the IRC foam recommendation are for R30 center cavity values, and you're going to be about 5/3 higher. If the foam were to be a single thickness rather than a taper you'd be advised to bump that R5 up to R8.4 (2" of Type-II EPS), but if you use cellulose as the cavity fill the moisture buffering capacity of the cellulose buys you sufficient resilience to get by with the thin edge at only R6.3 (1.5" of EPS, or 1" of polyiso)

    For the rationale on whole-assembly R values that high, read the whole first chapter, but note the R-values for climate zone 3 in Table 2 on p. 10 of this document:

  6. James Morgan | | #6

    In 1933 a group of European architects took a cruise around the Mediterranean and decided the whitewashed flat-roofed parapet-walled village buildings they saw there should be the model for a 'rational' style of architecture applicable anywhere, in any climate and in any culture, whatever local indigenous, traditional building wisdom may tell us to the contrary. We have suffered the consequences ever since.

  7. James Morgan | | #7

    Which is my way of saying, why not put a 'real' roof on your projects. Climate-appropriate overhangs and roof pitches designed to actually and reliably protect the vulnerable structure below. Oh, and by the way, there are many better ways to light artwork than ceiling cans. As any decent contemporary art gallery can attest.

  8. GBA Editor
    Martin Holladay | | #8

    Your story is poetic. I'm not sure whether it is an allegory or a historical anecdote, but I can see the scene in my mind's eye: the young architects, some in berets, speaking a mixture of French, German, and Italian, sipping their Chianti or absinthe as they lean over the ship's rail off the coast of Santorini, pointing to the cluster of homes on the hillside. A single donkey, led by a villager, makes its way up the lane to the top of the hill; the sun is low in the sky, and the weather is mild.

    It's enough to make a cold-climate roofer -- one who has suffered the consequences of their architectural epiphany -- a little bitter, isn't it?

  9. James Morgan | | #9

    Tom Wolfe's 'From Bauhaus to Our House' is a good read for a beleaguered cold-climate roofer on a winter's evening. May not make him feel better, but maybe that his grumbles are justified.

    Just for fun, here's a picture I took recently of a building in Los Angeles that's not thirty years old. Note the staining and discoloration of the stucco and the vertical 'accent' siding. Goodness knows what's going on behind that mess. The drapes are in tatters from excessive solar exposure. And what about those unreachable can lights! These buildings do not age gracefully, even in southern California's mild climate. Be sure your clients have deep pockets for the ongoing costs. As Mies did NOT say "Less is more ... trouble."

  10. James Morgan | | #10

    While I'm on the subject here's another tribute to haute modernisme. Los Angeles County Museum of Art is a collection of uninspired shoeboxes (mostly - I'll except the glorious and eclectic Japanese Pavilion) containing an extraordinary collection of beautiful artworks from all over the globe. And this minimalist triumph sits in the brutal and otherwise empty plaza out back. Yes, it's a very large rock. Sitting over a huge concrete trench. Countless carbon molecules were released into the atmosphere to achieve this statement. The guy on the left is there to tell people not to sit on the low concrete wall lining the trench, I guess so they don't mar the composition.

    OK I'll shut up now.

  11. GBA Editor
    Martin Holladay | | #11

    Thanks for the photo showing what happens when you design a cube without roof overhangs. I have my own collection of ugly no-overhang houses with siding problems, but most of them are prints, not digital photos, so they aren't easily accessible.

    How many U.S. cities are blighted by a concrete plaza without any human-scale features to invite people to congregate and relax? Dozens and dozens. When public spaces are degraded in this way, the community suffers for generations.

  12. Scott West | | #12

    Dana - Thanks for all the valuable input. I'll have to see if EPS is strong enough to carry the additional dead load of a roof deck that would be my only concern there.

    James, Martin - Ok, wow. I didn't imagine this thread degrading into a negative treatise on contemporary architecture. Some "real" input relative to my inquiry would have been appreciated instead. Good to see though that the tired stereotype of the elitist architect is alive and well. I didn't think since this site is called Green Building Advisor that it was the appropriate forum for discussions of personal esthetics but since you brought it up I'll respond with the following general thoughts:

    - I actually agree with many of your points about modernism ill-effects - especially in regard to public spaces. The idealism just did not take into account the realities of the human condition. But if you understand the historical context, you can understand more easily how it happened. At the turn of the century, most cities of any decent size were pretty dirty, polluted,overcrowded places so the idea of creating a decent standard of living through new ideas about mass production brought on my the industrial revolution seemed enticing. And the ideal of freeing the pedestrian space by building towers among large swaths of open , clean, public space seemed to make sense at the time.

    -Does holding up a few bad examples justify maligning an entire genre of architecture. One can find examples of bad design and lack of appropriate maintenance across ALL periods and styles. I'm sure I could send you dozens of photographs of the same issues with traditional homes. Once you go over one-story in height. most normally-sized overhangs don't provide that much protection to the lower floors anyway. How often does the rain fall perfectly vertically?

    - I always engage in passive solar design with software that analyzes sun angles, always provide overhangs where possible/needed, and attempt to protect any high-potential-weatherability materials with overhangs where I can.

    - Most problems I see with modern homes area result of trying to build them without a proper budget or without the guidance of an actual architect (or good builder). If you want something cheap and absolutely bulletproof then yes, you should stick with building a generic traditional home that's time-proven. But, you only live once and some people want to feel engaged and inspired by their living space and have the means to do so and are willing to live with some risk and some additional degree of maintenance.

    - I'm sorry that you have a personal vendetta against can lights but there's probably a reasonable and balanced way to incorporate these without "destroying" the energy-efficiency of a house. Yes, there are lighting methods that could be used but they are all generally "clutter-y" and do not provide an adequately clean look and are typically more expensive. Again, most of my client's goals are not to build the absolute most energy-efficient home possible but to find a balance between inspired architecture, clean esthetics, and energy-efficiency and sustainability.

    - I agree that we should not necessarily ignore the wealth of our combined historical building experience in regard to what works and what doesn't but at the same time are you saying that no one should ever try anything new? It's always puzzled me, for instance, why many people embrace new technologies when it come to things that go IN their house but not so much when it comes to outward esthetics. By the same token ,why aren't we all still driving around in Model T's? If no one ever tried anything new and different we'd still be living in the stone age.

  13. Dan Kolbert | | #13

    "I'm sorry that you have a personal vendetta against can lights but there's probably a reasonable and balanced way to incorporate these without "destroying" the energy-efficiency of a house."

    I'm not sure that's much of a response. You're taking perhaps the most important interior surface for thermal and moisture control and chopping a bunch of holes in it.

  14. Scott West | | #14

    If you have 4" of EPS or Polyiso on the roof deck plus 4" of cc foam on the underside of the deck then you have approx. R-44 and you still have 8" of free space throughout the ceiling plane for the cans, right? So there would be no thermal holes in this scenario. Obviously not as green a solution as cellulose but arguably workable right? And I suppose you could debate the merits/cost of filling the remaining space with cellulose despite the "thermal holes". Or forgo the underdeck foam and still dense pack the cellulose full-cavity wher you would have a continuous layer 6" deep and then another 6" deep layer with some thermal holes in it. Still a pretty good R-value all-in-all, right? Though not sure you can blow cellulose effectively around cans??

  15. James Morgan | | #15

    Sorry about the rant Scott, it was probably undeserved. But whatever happened to form follows function? It's such a shame really - the core insights of modernism, the blowing away of the cobweb of rote decorative detail and rigid classical formalism, at some point solidifying into a style and a brand, propagated by the architecture schools to be replicated worldwide, the fast food franchise of building design. Boy, there are some great, wonderful, glorious modernist buildings out there, but even the greats got it wrong as often as they succeeded, and as a commodity product for the journeyman architect the failure rate has been pitiful. If architects have a reputation for elitism and detachment from the real world of sky-high operational and maintenance costs, this is how they've earned it.

    By the way, modernism is not 'new', or even 'contemporary' any more. To follow your Model T metaphor, adherence to the modernist style - rather to its precepts - is like driving around in an Edsel, sixty years late. Nice looking car, if that's the look you like, but…….

  16. James Morgan | | #16

    And yes, due to the difficulty of creating an air barrier at the ceiling without access from above via an attic void I'd say your best bet, can lights or no can lights, would be foaming the underside of the roof deck. Not the greenest answer but it would seem to be what you're stuck with.

  17. GBA Editor
    Martin Holladay | | #17

    My comments on modernism were in response to James Morgan's posts. They were not directed at you. However, you have correctly guessed that I prefer architectural styles that include sloped roofs with generous roof overhangs. My statement of preference was not intended to impugn your designs in any way.

    You wrote, "Wow. I didn't imagine this thread degrading into a negative treatise on contemporary architecture. Some ‘real’ input relative to my inquiry would have been appreciated instead."

    Actually, Scott, I did my best to provide real input relative to your inquiry, in Comment #2, in which I advised that the best approach was to install rigid insulation above the roof sheathing or spray foam insulation below the roof sheathing. The main reason for this advice was the fact that the ceiling would include recessed can lights.

    You wrote, "Good to see though that the tired stereotype of the elitist architect is alive and well." I'm sorry if my light-hearted jest was offensive. I was hoping that my references to berets and absinthe made it clear that I was playing on tired stereotypes rather and commenting seriously on U.S. design professionals.

    You wrote, "I didn't think since this site is called Green Building Advisor that it was the appropriate forum for discussions of personal esthetics." Well, Scott, aesthetics are an integral part of residential design, so I would certainly hope that there is room on this site to discuss aesthetics. Your input and opinions are valuable, and we welcome all readers to join such discussions. I certainly appreciate your comments on the origins of modernism.

    You asked a rhetorical question: "Does holding up a few bad examples justify maligning an entire genre of architecture?" My answer is, "Of course not." Trust me, Scott -- here at GBA, we make fun of traditional architecture too, especially when errors result in catastrophic failure.

    You wrote, "I'm sure I could send you dozens of photographs of the same issues with traditional homes." Bring them on! I love failure stories. We all learn from disasters.

    You wrote, "I'm sorry that you have a personal vendetta against can lights but there's probably a reasonable and balanced way to incorporate these without destroying the energy-efficiency of a house." I'm not sure I would call it a vendetta, but it's true that I don't like can lights when they are installed in an insulated ceiling assembly. The best way to incorporate them is to separate the ceiling layer and the thermal boundary. For example, if you have a thick layer of rigid foam above the roof sheathing, it's perfectly reasonable to install recessed can lights -- because the can lights aren't interrupting the air barrier or the insulation layer.

  18. Expert Member
    Dana Dorsett | | #18

    Scott: You can get tapered EPS in several densities, but even Type-I EPS (1lb per cubic foot nominal density) is commonly used in commercial roofing apps. If you're putting a dynamic deck that you'll be walking on regularly it's probably worth springing for Type-II (1.5lb nominal), but it would be ridiculous to use Type-IX (2lb goods).

    Using four inches of closed cell foam blown with HFC245fa when you have the space for low-impact fiber is a crime against the planet, IMHO. With 8" of cellulose instead you'd still have 4" of space for electrical & mechanicals, and a HIGHER (and more stable!) total R-value. If you installed the recessed lighting cans and used air-tight insulation contact rated versions it would be even simpler/cheaper and you could give it the full-fill for even higher-R.

    A foot of 3lb cellulose even has substantial thermal mass- about 1BTU/degree-F per square foot of ceiling area, which is the equivalent of adding another inch of gypsum board, which has a measurable benefit on peak cooling loads.

    Polyurethane has a comparable specific heat per lb as paper/cellulose but at 4" of 2lb foam you're looking at about 0.7lbs per square foot, or less than a quarter the thermal mass of a full-fill of cellulose, and about half the R-value of a full fill of cellulose.

  19. J S | | #19

    In regards to the recessed cans, if they were IC rated, they would no longer be an issue right?

  20. GBA Editor
    Martin Holladay | | #20

    I.C.-rated cans still leak air, and they still displace insulation, thereby creating a hot spot in a location with thinner insulation than elsewhere. They also accelerate the stack effect, because the warm light bulbs pull air through the insulation, creating little chimneys.

  21. Scott West | | #21

    Sorry for going MIA recently. Working 16 hour days 7 days a week right now.

    Thanks for all the input.

    I'll look into rockwool as Dana mentioned but this probably can't be tapered for sloped roof right?

    There may be an esthetic limitation in many cases to how thick I can go with the above-deck rigid insulation since the whole roof edge becomes equivalently thicker. You have to have min. 1/4" per ft. of slope to roof drains so depending on spans that 4" average could mean up to 9 or 10" in thickness at maximum points. This also means that 2" average depth as advised by Dana, which would reduce the overall amount of less green-friendly EPS, may not be possible. Let's assume for moment then that I can only get average 4" depth tapered EPS on the roof. At R-vlaue of 4.0 per in. that's R-value of 16 for above deck. If I assume ceiling cans are 6" tall then that leave 6" of uninterrupted space into which cellulose could be packed. (R-value = +/-21, 3.5per in, right?). That's a decent but not amazing R-37 total. So, the question in this scenario is - is the extra 6" of cellulose to fill the remaining cavity full-depth still worth the extra expense despite the trade-off the thermal holes creted by the the an lights? Also, at what thickness does the cellulose require extra support so as not to deflect the 1/2" sheetrock ceiling and does this add significantly to the cost?

    Also, given that there's no airspace like a traditional roof I assume thermal bridging and general radiant enegry is more of a factor even with an all-white "cool roof". Is the addition of any kind of radiant barrier product above the deck worth the expense and effort or 12" of cellulose would be more than enough to offset any advantage there?



Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |