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
Q&A Spotlight

Pondering an Attic Conversion

Adding spray foam to the underside of the roof sheathing will turn a vented attic into a conditioned space — but will it also create problems?

Insulating the bottom of a roof deck will turn an attic into a conditioned space, but a homeowner whose house is about to undergo this transformation has some concerns.
Image Credit: Fine Homebuilding

Rob Graff is getting a new roof, and with it an opportunity to turn his vented attic into an insulated, conditioned space.

But he’s also got some concerns.

“They will be spraying 5 inches or so of closed-cell foam directly against the decking,” he writes in a Q&A post at GreenBuildingAdvisor. “This will fully seal my attic. My question is, how does this work if you do not have air conditioning in the house, other than window units in the bedrooms and one in the living room?”

Graff’s two-story house includes a partially finished attic, and he’s not sure how that can be vented once the roof deck is insulated.

“My concern is that, because heat rises, the attic will gather heat. How do I get that heat out of the attic if it has no vents? Also, how would I integrate a whole-house fan into this design?”

Graff’s questions are the topic for this Q&A Spotlight.

The attic is just another room

Once the roof deck is insulated, the attic becomes a conditioned space just like any other room in the house, GBA senior editor Martin Holladay says. “If you install closed-cell spray foam on the underside of your roof deck, creating an unvented attic, there is no reason to believe that your attic will be warmer than a vented attic,” Holladay writes. “If anything, it will probably be cooler.

“Your attic will be within 5 or 10 degrees of your living room temperature. If your living room is hot, your attic will be, too. If your living room is cool, your attic will be, too.”

The closed-cell foam, Holladay adds, should have only a marginal effect on how long the new roof shingles will last. Although the insulation will make the roof shingles “somewhat hotter” than they would be over an unconditioned attic, insulation is only one of a number of factors affecting shingle temperature.

“The most important factor affecting shingle temperature is shingle color,” Holladay says. “If you want cooler shingles, out of the belief that cooler shingles last longer — and they probably do — then install white shingles, not dark shingles.”

Incorporating whole-house ventilation

Graff currently places a fan in an attic window for nighttime cooling. With downstairs windows opened, and the fan in the attic window running, warm air is pushed out of the house and cool outdoor air is drawn in to replace it. This is an effective, low-cost way of cooling down a house without the use of air conditioning.

With an insulated roof deck, Graff wonders, what happens to this strategy? “So how does warm air leave the attic?” he asks. “It sounds as if the air in the attic will never get changed, and I will no longer be able to bring in cool air at night.”

Holladay noted that nighttime ventilation cooling — using either a conventional whole-house fan or the NightBreeze system — is a good way to cool a house in any climate that has cool tempertures at night. Graff responded that the NightBreeze system seems like something that would be added to a central air-conditioning system, not a stand-alone solution.

“When you convert a vented unconditioned attic into an unvented conditioned attic, you are incorporating the attic into your conditioned space,” Holladay replies. “It’s just like another room in your house. It shouldn’t be significantly warmer or colder than any other room. It’s just like a second-floor bedroom or a third-floor bedroom. But If you want nighttime ventilation cooling using a whole-house fan (usually installed in the ceiling of the upstairs hallway), it’s much easier to do with a vented, unconditioned attic.”

But, James Morgan writes, there’s no reason Graff wouldn’t be able to continue using an exhaust fan installed in his attic window — if he really needs to. “I see no reason why this arrangement will not continue to work in the new attic regime,” Morgan says. “To get the proper benefits, of course, there has to be an open air path to the attic space where the window is located and all window AC units need to be shut off.

“A whole-house fan is simply a bypass arrangement to enable nighttime cooling in homes without such a convenient layout as you currently have. It’s not so much that it won’t work, but that you don’t need it.”

Insulation plan has some disadvantages

Spraying 5 inches of foam on the bottom of the roof deck has at least a couple of problems, says Dana Dorsett.

Because closed-cell foam is vapor-impermeable, it would trap moisture at the roof deck, Dorsett says, and using foam at that thickness is a lost cause from an insulation point of view because of thermal bridging through the rafters.

“What’s more,” Dorsett says, “the blowing agents of most 2-lb. foam have a huge life cycle global warming hit, making anything beyond the first inch or two net negative for the climate.”

Instead, Dorsett recommends spraying 1 inch to 2 inches of foam against the roof deck and finishing with damp-spray cellulose or high-density fiberglass, a strategy that’s “the best of both worlds.” In support of his suggestion, Dorsett refers Graff to a Building Science Corporation report on the subject.

“All insulation systems are compromises,” Holladay says. Installing a continuous layer of rigid foam insulation over the roof sheathing or under the rafters would control thermal bridging, he advises.

“Unlike Dana Dorsett,” Holladay adds, “I don’t think that installing 5 or 6 inches of closed-cell spray foam on the underside of your roof sheathing will cause any moisture problems. It’s true that the roof sheathing won’t dry to interior to any degree — but as long as the roof sheathing is dry on the day that it is insulated, the system should perform as designed.”

Our expert’s opinion

Peter Yost, GBA’s technical director, adds these thoughts:

First, it is always helpful to know the climate that any assembly is facing, whether there is any HVAC equipment in the attic, and what the roof cladding is. I am going to assume that this is a cold climate, there is no HVAC equipment in the attic, and the roofing consists of 3-tab asphalt shingles.

Given these assumptions, here is my hygrothermal assessment (that is, my assessment of the energy and moisture performance of the attic space and roof):

1. Net energy efficiency gain by moving the air and thermal barrier to the roof line.

In almost all climates and building types, it is a net gain to “cathedralize” the attic, even if this space is not going to be used as fully conditioned space or living space. This is especially true for hot climates, but in many cases holds for cold climates as well, in large part because the roof plane is usually the easiest plane to get these two barriers continuous, and because the increased volume of the overall inside space of the building typically represents a relatively small energy penalty. Particularly in a cold climate, sealing air leaks at the eave is critical to combating ice dams.

2. Using the same attic window(s) to ventilate at night.

There is no reason that Rob Graff can’t continue to pull air from the living spaces below, up and through the cathedralized attic space. In fact, the significant moderation in temperatures afforded by the roofline air and thermal barriers will reduce the need for night-time cooling.

3. Impact on attic solar heat gain of roof cladding type and color.

The thermal performance of the roofing (its reflectance — the fraction of solar energy the roof cladding rejects — and its emittance — the fraction of absorbed energy the cladding can re-radiate) is important. But take a look at the numbers below for different roofing materials. White shingles are much better than black, but white metal roofing is far better than white shingles at rejecting the sun’s energy and shedding any that it absorbs. Interestingly, while Galvalume has a really high reflectance, the small portion that it absorbs is not easily re-radiated or emiited, so a white metal roof is far better at keeping an attic cool than a bare metal roof. We don’t know the age or condition of Rob’s roofing, but switching to white shingles is not nearly as effective as switching to white metal, albeit at a much higher initial cost.

  • White 3-tab asphalt shingles (representative numbers from Energy Star): Initial reflectance – 0.66; Initial emissivity – 0.85.
  • Black 3-tab asphalt shingles (representative numbers from Florida Solar Energy Center): Initial reflectance – 0.034; Initial emissivity – 0.90.
  • White metal roofing: Initial reflectance – 0.84; Initial emissivity – 0.91.
  • Galvalume metal roofing (bare metal): Initial reflectance – 0.90; Initial emissivity – 0.06.

4. Drying potential of the closed-cell spray-foamed roof assembly.

Given that the vapor permeability of both the spray foam and the asphalt roofing shingles is very low (both are Class I or Class II vapor retarders), the new roof assembly has little to no drying potential. While Martin Holladay is correct in saying that if the roof is dry at the time of spray foaming, it can stay dry, that is really dependent on just how much confidence Rob Graff has in the ability of his roof to shed bulk water and not leak. A very simple roof plane with few penetrations and in a sheltered location is one thing, but a complex roof plane with lots of penetrations in an exposed location is a whole other matter. I generally recommend strongly against assemblies without one dedicated direction of drying potential, unless the roof assembly has been designed and constructed (and will be maintained) flawlessly.

5. Need for ignition barrier on spray foam.

One code inspector may give a homeowner a pass for leaving the spray foam exposed in a non-living space;another may not. But frankly, if the pitch of the roof brings any utility to the newly enclosed attic space, Rob Graff will end up storing stuff or inviting people to this space, and it will need an ignition barrier, easily provided by gypsum board as the attic is finished.


  1. greenhouse437 | | #1

    insulated eaves
    Been thinking and reading about this for a few years. There seems to be the risk of homeowners thinking that if they insulate eaves and gable walls with 4-5 inches of insulation--plus even furring out from the eave joists with two by fours before installing sheetrock to create a thermal break--that that 'conditioned' attic space will then be a comfortable place to inhabit. So we're not talking about creating a habitable or occupiable space, which in many jurisdictions would require a permit, change of CofO, new emergency egress from attic, maybe a sprinkler system before you could then install heating and AC. In the purely 'conditioned' scenario, a room that is ten degrees below the heated rooms below in winter and ten degrees above the air conditioned rooms in summer won't be that comfortable. And if the homeowner does not remove the existing insulation under the attic floor, then the temperature variation will be greater. (A full-size attic walk up entry door being left open at times will certainly mitigate the floor insulation.) Certainly better for a central AC air handler to function in. There are also the potential issues raised by GBA of off-gassing of the spray foam. In our community we are allowed to create an unvented attic with finished sheetrock but only if that space continues to be used for storage only. We decided against it for now, until possibly a time when we can afford to make it a habitable space and afford the expected property tax increases, and a healthier spray foam comes on the market. For now we are filling insulation gaps under our attic floor as best we can.

  2. User avater
    Dana Dorsett | | #2

    5" of foam is NOT a Class-I vapor retarder.
    Tight though it may be, @ 5" most closed cell foam is still a bit north of 0.2 perms- a stiff class-II vapor retarder maybe, but still well out of the class-I range. ( Even at 8" thickness most are still above the 0.1 perm mark, which is the definition of the Class-I boundary.) That's not to say that I'm advocating the approach.

    The water-blown Icycene MD-R-200 product would still be a minimal class-II vapor retarder @ 5", barely ducking under 1 perm at that thickness, which may be an acceptable balance on moisture performance, though more expensive than the flash-inch or two of a lower-perm closed cell foam (which also works.)

  3. Michael Blasnik | | #3

    conditioned attic efficiency claim
    "In almost all climates and building types, it is a net gain to “cathedralize” the attic, even if this space is not going to be used as fully conditioned space or living space"

    Can you provide any evidence to support this claim? It seems hard to make the math work on this given your stated assumptions of a cold climate with no HVAC in the attic. The conditioned attic has greater area and lower R value (at least for this 5 inches of foam example) than an insulated ceiling. Most attics can be air sealed at the ceiling level fairly well so air leakage differences will not usually be able to make up the difference. Conditioning an attic will usually increase HVAC energy use unless there is HVAC in the attic and/or large and unsealable air leakage problems (dozens of recessed can lights?). I'd be interested to see any studies (or even a reasonable calculations) that show a net efficiency gain for conditioning attics that don't have HVAC equipment.

  4. User avater
    James Morgan | | #4

    I agree with Michael.
    Cathedralized insulation obviously makes sense for the room-in-the-roof construction indicated here and for retrofitting to existing attic mechanical spaces, but extending this generally to non-functional attics as Peter suggests is questionable and misleading on several counts. Michael's points are worth repeating:

    First, R-values matter. Vented attics with raised-heel rafters or trusses can go to R-50 or R-60 simply and economically, whereas rafter-level insulation seldom seems to go beyond R-30. 'Because air-sealing' is the traditional poor excuse for low SF R-values. Yes, we need good workmanship and attention to detail to properly seal attic floors - no less do we need it to spray foam safely and reliably on the underside of a roof deck.

    Secondly, surface areas matter. Peter is correct that the additional volume of the conditioned attic plays a negligible role in heat transfer conditions: in fact it could be argued that it plays no role at all in steady-state conditions. What does make a difference is the significant increase in surface area represented by insulating a sloped roof. With a moderate slope, say 9/12 or so, the affected area including gables will be 50% larger than the equivalent flat ceiling of a vented attic. For a given functional volume would you rather have a thousand square feet of enclosure at R50, or fifteen hundred square feet at R30?

    Finally, and I'm genuinely curious about this, what numbers should you use for the Delta-T on a cathedralized attic assembly: external air temp. or the surface temperature of the roof deck? Depending on roof construction, seems to me the deck can be far hotter than both external ambient and the air in a conventional vented attic, and that's the number you should use for the heat transfer calc.

  5. User avater GBA Editor
    Peter Yost | | #5

    Technical comments: vapor retarder class and thermal performance
    Good points on vapor retarder class of closed cell spray foam and my generalized thermal performance characterization of cathedralized attics across climates and uses.

    Stiff class II retarder is correct for 5 or more inches of closed cell spray foam. Using the reciprocal rule of thumb (the vapor permeability of materials like spray foam decreases with the reciprocal of the thickness), if 1 inch of closed cell is about 1 - 2 perms, then 5 inches would be around 0.2 to o.4.

    My claim regarding the overall better thermal performance of cathedralized attics versus vented attic at the ceiling was based on several studies BSC and other Buiding America teams have conducted over the years ( I agree that the generalization to all climates and all conditions of use I made it too strong; it focuses on the fact that achieving continuous air and thermal barriers at the roof line is generally much easier than at the ceiling and is admittedly heavily influenced by the typical presence in many homes of HVAC in the attic space.

    In terms of what temp to use to get the delta-T: probably need to use hourly modeling tools to reflect how much both ambient and surface temperatures swing diurnally.

    Thanks for your added correcting and clarifying comments.

  6. John Brooks | | #6

    I Agree with Michael and James
    What's not to like about a Vented Attic?

  7. User avater GBA Editor
    Martin Holladay | | #7

    Response to Peter Yost
    Thanks for your comments. I have edited the text of the article slightly to reflect your corrections.

  8. User avater
    Dana Dorsett | | #8

    You underlined the wrong bits (response to John Brooks)
    The last sentence is critical:

    "But for all kinds of reason we manage to screw it up."

    To which I'd add "And HOW!!".

    This project is a retrofit. In new construction with well trained crews it's possible to get the proper level of air sealing at the attic floor plane, and provide sufficient space at the eaves for high-R out over the top of the wall while still getting the requisite amount of ventilation air. But tin a retrofit, it may be impossible to achieve the necessary R-value near the edges, and air sealing the attic floor plane can be extremely labor intensive to achieve. (In snowy parts of climate zones 4 & 5 it takes R50+ to eliminate ice dams, and thin spots in the at the top of exterior walls MATTER.) It's almost always going to be easier to air-seal at the roof deck rather than the attic floor (even in new construction), and blocking the stack effect infiltration has a disproportionate effect on energy use beyond the R-values installed.

    In most retrofits you tend to do much better on energy performance insulating at the roof deck, even without mechanicals in the attic. That is small-t "true", even though on a purely theoretical U-factor x delta-T x area basis the conducted heat losses/gains are higher. (No published supporting data here, just mounds & mounds of anecdotal evidence, plus direct experience. :-) ) I'm sure there are many exceptions to prove the rule, but I share Peter Yost's perspective on that, based on projects I've done (and in the case of my own home, re-done.)

  9. Harry Voorhees | | #9

    vented roof with spray foam?
    Would Dana's suggested flash and fill approach, or 5" of icynene against the roof sheathing, provide sufficient drying potential for a mild roof leak?

    I'm planning to condition the attic of my c. 1890 house in Andover, MA which has a cross-gabled roof with central beadboard finished interior rooms with knee walls. I've been leaning towards maintaining a vented roof in case a leak should occur (e.g., among the 100+ fasteners securing a solar PV array to the roof).

    One option is to insulate the room envelope - knee walls, slopes (using baffles to leave 1-2" air gap under the sheathing) and attic ceiling. But providing an air barrier at the intersections between the knee wall, roof, floor, and knee wall doors seems tricky.

    Including the cubbies and dormers in the conditioned space would seem to be more effective. Could the roof from the eaves to above ceiling be baffled, then sprayed with foam? That way only the rafter bays abutting the dormer valleys would be unvented.

  10. User avater GBA Editor
    Martin Holladay | | #10

    Response to Harry Voorhees
    Q. "Could the roof from the eaves to above ceiling be baffled, then sprayed with foam?"

    A. Yes. More information in the following two articles:

    Two ways to insulate attic kneewalls

    Creating a Conditioned Attic
    (In this article, I wrote, "Even if you insulate between your rafters with an air-impermeable insulation like spray polyurethane foam, you may want to provide a ventilation channel under your roof sheathing. The main function of such a ventilation channel is to separate the roof sheathing from the foam; this facilitates future repairs of sheathing rot.")

  11. Josh Rountree | | #11

    Video - Building Science Fundamentals: Roof, Part 1: Ventilation
    I'd like to echo the same thing Dana Dorsett did.

    In Joe's video he says:
    "But for all kinds of reasons we manage to screw it up. The single most important thing you have to remember about a vented attic is that ceiling plane needs to be air tight."

    The part I'm more concerned with in an unvented attic is the underlayment used between the roof material and the roof deck and the drying potential of the roof deck - see #4 from Peter Yost above. I see the details needed to get that right more costly than providng an air space, i.e. venting the roof deck. I've seen several people mention the need for a vapor barrier as underlayment instead of 30 lb felt, but I don't see this need emphasized. It seems to me that if you don't provide an air space, or don't use a vapor barrier, it's a recipe for the roof deck to rot.

  12. Roy Crawford | | #12

    Ductwork in attic
    The benefits of an insulated roof deck vs. insulated ceiling has a lot to do with duct location. If the ducts and air handler are in the attic, insulating the roof deck essentially reduces duct losses to zero, which is significant, especially for leaky ductwork.

Log in or create an account to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |