GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Plus Icon Minus Icon Picture icon Hamburger Icon Close Icon Sorted

Community and Q&A

How best to insulate a 12″ rafter-framed cathedral ceiling?

cconti | Posted in Green Building Techniques on

Hi all,

After reading (thoroughly, several times!) Martin’s article, “How to Build an Insulated Cathedral Ceiling,” I still have a number of questions.

First, my situation:
I’m building a garage with loft living space in central Maine (climate zone 6). The cross-section is a cathedral ceiling with rafter ties (image attached). The roof assembly is 12″ rafters, synthetic roof underlayment (a product called “Elephant Skin” by Warrior Roofing) and screw-down metal panels with a ridge vent. All of that is already done. The eave fascia is not yet completed; the existing plan is for soffit vents.

I had planned on insulating with fiberglass batts in the rafter cavities (vented by vent baffles) plus a layer of polyiso underneath the rafters, but because of the problems that that poses regarding nailing surface for electrical junction boxes, I’ve abandoned that plan.

My reading of code suggests that minimum ceiling R-value in my area is R-49 (please correct me if I’m wrong there).

I got in touch with a local spray foam contractor who recommended 5″ of foam sprayed to the underside of the sheathing. But as far as I know, 2lb. foam is only R5-R6 per inch, meaning 5″ would only get me to R25-R30, which is well below even the code minimum. I have an email in to him asking about this (in our conversations he didn’t say anything about code minimums or about including any other insulation in addition to the foam… so am I misunderstanding something? Or was this guy really going to leave me with less insulation than required, and not mention anything about it???), and I’ll update this post when I hear back from him.

After reading Martin’s article on insulated cathedral ceilings above, I’m unclear on a few things. I know that if you install foam on top of the roof sheathing that there needs to be at least R-25 (in my area) of it in order to keep the sheathing above the dew point and therefore prevent condensation that will rot the sheathing. But the article was less clear about foam sprayed underneath the sheathing. I think I understood that the required R-value is the same (R-25, in my area), but I’d love clarification of that.

Second is the question of venting. I understand from the article that using an air-impermeable insulation like spray foam allows you to use an unvented attic approach. I have a couple of questions about this though. First, my lumberyard specified (and my builder used) the “Elephant Skin” underlayment mentioned above. I was unable to find any vapor permeability information about that product, so I called the manufacturer and reached an individual who didn’t really seem to have any specific info but said he would not recommend using spray foam underneath sheathing with the Elephant Skin applied on top of it. What are people’s thoughts on that? I’m a bit nervous about sandwiching the sheathing between two different vapor-impermeable layers which would seem to prevent it from breathing, but I seem to remember reading where Martin said that wouldn’t be a problem. So I’d love clarification on that point.

Finally, regarding the total R-value of the assembly:
If I have 5″ of foam applied to the inside of the rafter cavities, do I then fill the rest of the cavity with fiberglass/rock wool to maximize my total R-value, essentially using a flash-and-batt method? Are there any problems with this?

(This is the first time I’ve built a building from scratch, and while I had thought I’d planned the build sufficiently in advance, it is dawning on me more and more that that is not the case. It seems just about the only thing I’ve done right is to avoid using recessed lighting cans, instead going with shallow boxes and those awesome new Philips flush-mount LED fixtures…)

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. cconti | | #1

    Thank you all in advance for any advice you can give... this has definitely been a learning experience for me. :-/

  2. Expert Member
    Dana Dorsett | | #2

    In zone 6 to meet the letter of code you'd need R49 center-cavity, and in an unvented assembly half the R-value would have to be air-impermeable foam in direct contact with the under side of the roof deck to be able to use a class-III vapor retarder on the interior. But in practice the roof deck will be just fine with 2" of closed cell foam (R12-R13) under the roof deck, with the rest being fiber insulation. To keep moisture from reaching mold-inducing levels in the rafter bays you can use an interior side "smart" vapor retarder such as Intello Plus or Certainteed MemBrain behind the gypsum board, detailed as an air barrier. (MemBrain is the cheaper option, and now even being sold through some box store outlets online, but I've yet to see it on box-store shelves. It's only 2-mils thick, but it's nylon- it's tougher than it looks, but still possible to tear. )

    A milled 2x 12 is 11.5", so if you installed 2" of cc foam you'd have 9.5" of space for fiber. If you compress a high density "cathederal ceiling" R38 batt into that space it will be 3/4" thinner than it's manufactured loft, but would still be delivering at least the R36 performance (or close enough to it) that it would still make code min. If you wanted to be absolutely sure, cut some rafter edge strips out of 3/4" thick rigid polyiso so that the 10.25" thick R38C batt just fits and you'd have a center-cavity R-value of R50-R51.

    At 2" of thickness any 2lb density closed cell foam is reliably in the middle of the Class-II vapor retardency range, at 0.4-0.7 perms and will not form a moisture trap with an impermeable layer above the roof deck, but will not have moisture accumulation rates in winter high enough to create a problem in the roof deck. MemBrain is also a class-II vapor retarder when the air next to it has a relative humidity below ~40%, as it would be all winter long, which limits the amount of moisture that accumulates in the fiber layer. But if the moisture content ever reaches mold-inducing levels it becomes more vapor-permeable than standard latex paint, so it does not inhibit drying toward the interior. As long as the MemBrain is reasonably air tight (which requires some detailing around the electrical penetrations etc.) the protection level is VERY good.

  3. GBA Editor
    Martin Holladay | | #3

    Q. "Or was this guy really going to leave me with less insulation than required, and not mention anything about it?"

    A. Yes. For more information on this topic, see It’s OK to Skimp On Insulation, Icynene Says.

  4. cconti | | #4

    Hi Dana,

    Thank you for your response.
    I'm having a little bit of trouble understanding: is the goal to minimize the vapor getting into the rafter cavity, or to maximize the vapor permeability out of the cavity into the living space (or both!)?

    I think this is a case of you solving the problem before I even know what the problem is!

  5. GBA Editor
    Martin Holladay | | #5

    Q. "The manufacturer ...said he would not recommend using spray foam underneath sheathing with the Elephant Skin applied on top of it."

    A. That is correct. That limitation is typical for most synthetic roofing underlayments. The limitation is clearly stated in the Elephant Skin installation instructions, which state, "ELEPHANT SKIN must be installed on a well ventilated roof with a pitch of 3/12 or greater."

    Is this limitation logical? Probably not for roofs with asphalt shingles -- but certainly for roofs with concrete tile roofing, cedar shingle roofing, or slate roofing. Does the manufacturer care if the limitation is logical? No. But if you don't follow installation instructions, you get no support or warranty service.

    The moral of the story: if you are building an unvented roof, use asphalt felt, not synthetic roofing underlayment.

  6. GBA Editor
    Martin Holladay | | #6

    Q. "The article was less clear about foam sprayed underneath the sheathing. I think I understood that the required R-value is the same (R-25, in my area), but I'd love clarification of that."

    A. As you guessed, the requirement for the minimum R-value of spray foam insulation under the roof sheathing is the same as the requirement for the minimum R-value of rigid foam above the roof sheathing.

  7. cconti | | #7


    "That is correct. That limitation is typical for most synthetic roofing underlayments. [...] Moral of the story: if you are building an unvented roof, use asphalt felt, not synthetic roofing underlayment."

    Eek. Well, given that the roof is already done and the synthetic is already on there (with metal roofing on top of it), what do I do now?

    Am I now limited only to a vented, fiber-insulated roof?

    (For what it's worth, I don't care about support or warranty service from the underlayment manufacturer, I just care about what's going to be best for the building!)

  8. Expert Member

    I don't want to bring up things that may cause you anxiety when they are probably not a problem, but my understanding is that the 2012 IRC says that rafter ties have to be in the bottom third of the span from plate to ridge. Your building section appears to show them in the top third where collar ties are usually located.

  9. cconti | | #9


    Sorry, you're right. Those are collar ties, not rafter ties. This structure doesn't have rafter ties per se.

    (I was a little bit leery about the design for that reason, but the building inspector signed off on it.... I guess in a few years we'll see if the ridge starts to sag... :-/ )

  10. GBA Editor
    Martin Holladay | | #10

    The advantage of roofing underlayment with variable permeance (for example, asphalt felt) is that the underlayment allows damp roof sheathing to dry to the exterior when necessary. In the case of metal roofing, drying to the exterior is possible if the roofing is installed on 1x4 or 2x4 purlins. Some configurations of metal roofing -- for example, traditional corrugated metal roofing with high ribs -- can also dry to the exterior.

    If you have installed standing seam roofing without purlins, then you won't get any drying to the exterior. In that case, your synthetic roofing underlayment will do no harm.

  11. Expert Member
    Dana Dorsett | | #11

    "I'm having a little bit of trouble understanding: is the goal to minimize the vapor getting into the rafter cavity, or to maximize the vapor permeability out of the cavity into the living space (or both!)?"

    There are two goals here:

    1: Prevent the wintertime moisture build up in the cold roof decking from reaching mold/rot levels. Moisture will build up in the roof deck over a winter because it's average temperature in winter is below the dew point of the interior conditioned space air, so whether by vapor diffusion through the materials or air leaks, some moisture will accrue until it's warmed up in spring. But a true vapor barrier would create a moisture trap- any moisture that gets into the roof deck could never leave.

    2: Prevent wintertime moisture build up in the fiber insulation layers to levels that would promote mold.

    The vapor retardency of 2" of closed cell foam is low enough to prevent roof rot, but not so low as to create a moisture trap, so it reliably deals with problem # 1.

    But the R-value of the foam as a fraction of the total R isn't sufficient to keep the surface of the foam above the dew point of the interior air (you'd need nearly 2x as much foam to get there, which is expensive and not very green due to the environmental damage of the blowing agents and polymer production), but the vapor retardency of the MemBrain is sufficient to prevent wet fiber insulation & mold levels unless there are big air leaks, taking care of problem #2. But if an air leak were big enough to reach mold-supporting levels toward the end of the season, the MemBrain becomes vapor open, promoting rapid drying. As long as it can dry before the TEMPERATURE is high enough to support mold growth, the problem is prevented.

    This strategy works remarkably, well, as long as you have enough foam-R that the average winter temperature at the interior surface of the foam isn't TOO cold, so that the drying season starts early, and the accumulating season starts late. At a 25% foam-R/total-R ratio (R13 out of R51 total) it cuts the accumulation season by something like 2 months in a zone 5 climate- it's a bit like moving it 800-1000 miles south from a temperature average perspective at the foam/fiber interface, but since climate zone 5 interior air is drier than it would be in that more southerly location, from a moisture point of view it's a bit like moving it to a warm temperate zone desert, which gives a smart vapor retarder ample room to work it's magic.

  12. cconti | | #12

    Hi Dana,

    Thank you for trying to help but I unfortunately didn't follow that at all. :-( It sounds pretty complex and very tricky to get right.

    I have another idea: use rafter vent baffles, and spray foam underneath them. Could this work? I posted another thread on that here.

  13. Dana1 | | #13

    Not tricky at all to get right. The abbreviated version:

    The stackup:

    * Roofing

    * Underlayment

    * Roof deck

    * 2" of 2lb density polyurethane directly on the roof deck

    * R38 high density batts

    * Smart vapor retarder (I just found out that some Lowes stores in MA now have MemBrain on the shelf, not sure about ME. )

    * Ceiling gypsum

    The foam protects the roof deck from rot, the smart vapor retarder protects the batts from mold.

    It works.

  14. Expert Member
    Michael Maines | | #14

    Chris, I'll reiterate Malcolm's concern. Unless your walls are balloon framed, with the collar ties located where they are, your walls WILL spread. The roof probably won't collapse but I have repaired/reinforced several structures with similar details and have seen many more. What happens is the collar ties put so much stress on the rafters, trying to keep the eaves from spreading, that the rafters deflect several inches at the collar ties. I've seen as much as 4" out of plumb walls with a 2" kink in the rafters. While you are still in construction mode, I would reinforce the walls somehow.

    Options include installing a few LVL columns balloon-frame style (continuous from the mud sill to the top plates); creating shear walls perpendicular to the exterior walls, connected to reinforced floor joists; installing a structural ridge; or using custom-fabricated metal brackets at the eaves with reinforced studs and rafters. If you balloon framed the walls, they will still spread but not as much. Trust me, it's a lot easier to deal with it now than it will be later. The building inspector holds no liability when your structure fails so don't put faith in his engineering abilities.

    As for your insulation issues, the other guys covered it pretty well. Personally I would furr the rafters down so can vent the roof and achieve at least R-49 with dense-packed cellulose, with a vapor-variable air barrier on the inside. You give up a little headroom but the short term and long term costs to you and the environment make it a worthwhile tradeoff.

  15. GBA Editor
    Martin Holladay | | #15

    Michael's right. This type of building needs continuous studs -- say, 13 feet or 14 feet long. I just built a garage two years ago similar to yours, and that's the way I framed it.

  16. KeithH | | #16

    Ok, you got tons of awesome professional advice. Here's aDIYer answer based on what I did to deal with an unvented vapor impermeable roof (EPDM) with inadequate insulation. Fill available space with roxul ComfortBatts. 2 R-23 batts will just fit inside a 2x12 rafter for R-46. If you truly need R-49 that won't work but if that is just a target, R-46 is close. Then install a sheet membrane for air tightness. As Dana mentioned, use a smart vapor retarder. Dana mentioned Certainteed Membrain; I haven't used that. I've used Proclima Intello Plus and really like it. Skip the regular Intello (non plus); I've had stretch on install and then recall over time breaking tape connections. The plus doesn't stretch or 'shrink'. You are relying on the membrane to eliminate moisture reaching your uninsulated sheathing so you'll need a well-detailed air-tight installation. I bet this idea is unpopular with the pros due to concerns related to vapor penetration. It does allow your sheathing to dry to the inside, which closed cell foam doesn't allow. But it is probably a bad idea in your climate versus my more desert climate (zone 5b). So I wouldn't recommend that for your climate but ...

    Another option would be flash and fill. Dana discusses this as well. I just want to mention that Roxul batts could work for this as well; flash fill to ~4" with cc foam (R-24), batt fill with a Roxul 2x8 (7.25"~ thickness) batts (R-30) for an R-54 assembly. For 2" flash (R-12), you could stack up a 2x6 (R-23) and 2x4 (R-15) ComfortBatt for an R-50 assembly. Be aware that the roxul is more pressure resilient so a proud installation of roxul will require a sturdy squish layer (membrane, drywall) compared to the exceedingly flattenable fiberglass. No big deal.

    Hope that helps.

  17. Expert Member
    Dana Dorsett | | #17

    Regarding Keith's commentary:

    Intello (even Plus) on it's own won't cut it in a zone 6A climate, not even close. I'm not convinced that it's truly sufficient protection in zone 5B, though it's a lot better than relying only on latex paint.

    At 4", a closed cell foam is hardly a "flash", in fact it has to be installed in two passes of 2", separated in time by a curing/cooling period to be done safely and without shrinking.

    At 4" you have sufficient foam to add another R25 underneath it without even the need of smart vapor retarders, and it adds significant structural strength to the assembly (more relevant in hurricane zones than central Maine), but the "extra" two inches of the second pass delivers far more lifecycle damage and is far costlier than a layer of 2-mil nylon (MemBrain). The second pass 2" of foam runs about $2 per square foot compared to about 50 cents per square foot for MemBrain + 2-3" of fiber insulation (installed costs).

    Just 2" of closed cell foam on the underside of the roof deck is sufficient to protect the roof deck, but it takes the smart vapor retarder on the interior side to keep the fiber insulation sufficiently dry when there is only 2" of foam at the roof deck.

  18. cconti | | #18

    Regarding the structural issue:

    Michael, thank you very much for voicing your concerns. When I was sketching the design for this building I was concerned the lack of rafter ties could result in the exterior walls bowing out, leading to roof sag. I relied on the building inspector's judgement that it wouldn't be a problem. That may have been a mistake.

    Of all the options you listed for ways to address the problem, the shear walls sound the most attractive, because we actually already have interior walls perpendicular to the exterior side walls (see the floor plan of the second floor below).

    The existing exterior walls aren't centered in the middle of the exterior side walls but instead are about one third of their length so aren't ideal from a structural standpoint, but hopefully they'll be sufficient.

    I can use structural plywood or sheathing to reinforce the walls, then use hurricane ties or similar Strong Ties to tie the shear walls into the side walls, then bolt the shear walls into the floor joists below (which are 16" I-joists). For additional strength I suppose I could also tie the joists in question to the adjacent joists with blocking or some similar method in the joist cavity.

    Would that be sufficient?

    Thank you very much for your help.

  19. Expert Member
    MALCOLM TAYLOR | | #19

    Unfortunately, despite your inspector's approval, your design is outside of the code requirements. That means potential fixes need to be engineered. Would your DIY solution work? Perhaps, but you are introducing a whole new set of structural forces into the building that to be blunt you don't understand. You need professional advice.

  20. user-4524083 | | #20

    Chris - Does the garage portion have a center beam and posts supporting the floor joists? If so,you could perhaps reconfigure the walls upstairs to contain posts over the posts down in the garage. If there are enough of these posts downstairs, your additional posts upstairs would transfer the load down to your slab, and prevent the ridge from deflecting. This assumes the slab could take the extra load and be reinforced. I've built dormered cape houses this way, with the ridge supported by posts that transfer down to a reinforced, thickened slab. Dropping those collar ties down another foot, while not to code, will help as well. I'm no engineer but just "eyeballing" it, the currently planned walls upstairs don't seem adequate even if bolted and stiffened. Good luck with this project. I think Michaels advice is right - fix it now before snow like last year hits us. I had 4' on parts of my roof in Brunswick.

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |