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

Untangling a Roof Insulation Goof

The layer of exterior insulation is too thin. Now what?

An under-insulated roof: A section of flat roof on this Climate Zone 4 house has a layer of rigid foam above the roof sheathing. But the homeowner will have a tough time getting the roof assembly up to the R-49 required by code. (Image credit: Ryan Lewis)

This Q&A Spotlight begins with a confession of sorts from Ryan Lewis, who thinks he may have botched an insulation job.

“Let’s suppose you screw up insulating a flat roof from the exterior in Climate Zone 4A,” he begins in a post at the Q&A forum.

The code minimum for attic insulation in this climate zone is R-49. Lewis’ roofing contractor has added 1 1/2 inches of polyisocyanurate foam insulation on top of the roof deck, which Lewis assumes will perform at its nominal R-value of 9.75.

If he adds enough fluffy insulation between the rafters to get the total roof  R-value to 49, Lewis runs the risk of creating moisture problems on the underside of the roof deck. Why? The roof sheathing will be cold, cold enough to become a condensing surface for moisture working its way through the cavity insulation. (For more information on this issue, see “How to Install Rigid Foam On Top of Roof Sheathing” and “Combining Exterior Rigid Foam With Fluffy Insulation.”)

If the layer of foam is too thin, can’t he simply add more to keep the roof sheathing above the dew point? Not now. The roof is loaded with solar panels. In order to add more foam, Lewis would have to remove the panels and the racks as well as the roofing.

What Lewis wants to know now is the maximum amount of fluffy insulation he can install between the rafters below the roof deck without incurring moisture related problems.

Balancing interior and exterior insulation levels

The issue is the proportion of interior to exterior insulation. In Climate Zone 4A, the International Residential Code prescribes R-15 to the exterior, and a total R-value of 49, writes Dana Dorsett. About 30% of the total R-value should be to the exterior of the fluffy insulation on the interior.

With R-9.75 on the exterior, he continues, the most cavity insulation he can add is R-23, for a total center-of-cavity R-value of 33.

With a low-slope roof, something less than 2:12, there will be less convection cooling in the roof and the roof deck would remain somewhat warmer, Dorsett adds.

“I’d be comfortable painting the underside of the roof deck with half-perm latex paint and installing R-23 rock wool or 1.8 lb. [per cubic foot] blown fiberglass under that,” Dorsett says. Computer modeling of the assembly with a program called WUFI could help allay concerns Lewis might still have.

“If there isn’t going to be any ceiling gypsum, stapling ~5-perm perforated aluminized fabric type radiant barrier would hold the batts in place and slow the rate of moisture accumulation at the paint/fiber boundary,” Dorsett says. “If there is going to be gypsum board on the underside, put the half perm paint on the gypsum, not the roof deck.”

The risk of overstating polyiso’s performance

Referring Lewis to the applicable sections of the IRC (here and here), Michael Maines mostly agrees with Dorsett’s assessment. Lewis needs at least 31% of the total insulation value in the roof on the exterior side of the sheathing.

If he uses the nominal R-value for polyiso (R-6.5 per inch), the maximum total insulation in the roof will be R-31.8, leaving about R-22 for the fluffy (air-permeable) insulation between rafters.

“But,” he adds, “since we are talking about the serious issue of moisture accumulation (and resulting damage) if it’s not done correctly, I don’t know why you would use a false value for polyiso. R-5.0 to R-5.5 per inch are more reasonable values for polyiso a few years old in cold weather. I’d use a value of R-8 or so for the polyiso, putting the total allowable fluffy stuff at R-18.”

Consider adding an interior layer of spray foam

These options leave the roof well below code-prescribed minimums. If Lewis wants to get closer to what it should be, Dorsett suggests a 2-inch layer of closed-cell polyurethane foam (with an HFO blowing agent) on the underside of the roof deck.

The spray foam would add up to about R-14, and with R-15 rock wool batts on the inside, the roof would approach R-39 at the center of the rafter bay.

Yes, adds GBA Editor Martin Holladay, this is one situation where Lewis might consider the “dreaded ‘foam sandwich.’ ” Dreaded because an assembly that traps a layer of wood sheathing between two impermeable layers of foam would seem to be begging for trouble. Should there be a leak, how can the roof dry?

“If the roof sheathing is dry,” Holladay says, “you can safely install closed-cell foam on the interior — and in that way you could make the foam portion of your roof assembly thicker (by adding the R-value of the exterior rigid foam to the R-value of the interior closed-cell spray foam — and using that R-value to calculate your permissible fluffy layer).”

Other risk factors for moisture damage

Jon R writes that Lewis should consider a variety of factors in calculating the potential for moisture problems. His list includes:

  • Air sealing the interior side.
  • The use of a “smart” vapor retarder, such as Intello membrane.
  • The choice of what type of fluffy insulation to use (cellulose is the better route).
  • What indoor humidity levels are likely to be (the drier the better).
  • The impact of the color of the roofing (darker colors absorb more heat from the sun).
  • The vapor permeance of the exterior.
  • What kind of roof sheathing Lewis uses (plywood is a better choice than OSB).

Reservations about spray foam

Lewis would rather avoid spray foam. “I’m generally pretty spooked about spraying foam,” he replies. “So living with below code R-values isn’t the end of the world.”

With full-width 2×6 rafters (the house is 87 years old), and using non-foam insulation, the most Lewis thinks he should install below the roof deck is R-24. Choosing R-22 or R-23 rock wool gets him pretty close.

If spray foam makes Lewis nervous, rigid foam does not. With that in mind, he wonders whether he could use 2 inches of polyiso below the roof deck (two 1-inch-thick boards with seams staggered).

“If you take the rule bending to the limit,” he asks, “why not just fill all of the cavities with rigid foam? Then the ratio constraint essentially doesn’t apply?”

What Lewis is suggesting, Holladay replies, is called the “cut-and-cobble approach.”

“When that approach is used in unvented roof assemblies, it’s associated with moisture accumulation and rot,” he says, “because it’s hard to keep all the seams airtight with changes in humidity and temperature and with changes in snow loads on the roof.”

Of course, should Lewis get beyond his reluctance to use spray foam, filling the entire rafter bay would solve a host of performance issues.

Six inches of foam would get the roof to roughly R-50, Dorsett says, although it would prevent any drying toward the interior — and the cost could be in the $8-a-square-foot ballpark.

“For the money,” Dorsett writes, “it’s hard to make a financial case for anything more than what’s needed for dew point control. At 2 inches, all closed-cell HFO blown foam is under 1 perm, a class II vapor retarder, but at the vapor-open end of class II, which is exactly where you want it to be for protecting the roof deck from winter accumulation without blocking drying too much.”

And, adding a 2-inch layer of foam should cost less than $3 per square foot.

Our expert’s opinion

Peter Yost, GBA’s technical director, adds this:

Once again, I am truly impressed with the level of the GBA exchange on this Q&A. I particularly like the list of factors by Jon R. Here are some additional thoughts:

Air sealing rules: vapor flow considerations are just not as important as achieving a sustained air seal.

Use of Intello Plus: After looking at technical documents from 475 High Performance Building Supply, I think that using a smart vapor retarder in a “middle” climate with significant heating and cooling needs would require checking in with 475 tech support. You can see those documents here and here.

Cellulose as the preferred roof cavity insulation: This is an air-permeable insulation that Building Science Corp. chose for its vapor diffusion port research (see JLC’s “Avoiding Wet Roofs – Part II). I want to see the final report on this for climate zones above 3 before giving cellulose the edge.

Indoor humidity levels: This is often a forgotten determinant of the evaluation of moisture performance of assemblies — nice catch.

Impact of roof “color:” Another good catch. But a better expression of this phenomenon might be highly reflective vs. heat-absorbing roof claddings. It’s not just about color but the idea is that “cool roof” claddings mean a big reduction in how much heat a roof assembly takes on, with heat being a powerful roof-drying mechanism.

Vapor permeance of the exterior: It has always puzzled me that the building code ignores this part of roof assembly drying performance. Yes, most roof claddings are class I or II vapor retarders, but not all.

Type of roof sheathing: Another good catch, and I would add roof sheathing boards. We have at least one local high-performance builder who uses roof and wall sheathing boards because of how much more forgiving they are in terms of moisture.

One factor that I would add to Jon R’s list: roof complexity. How worried I am about designed moisture performance versus actual installed performance is really driven by how simple or complex the roof is. While the final total R-value of the roof assembly may be primarily driven by cost, the one element of the roof assembly that can’t be compromised or determined by trade-offs is the airtightness of the roof assembly.


  1. Charlie Sullivan | | #1

    The roof color matters less for the part that is shaded by the solar panels. The solar panel shading will reduce the roof temperature on sunny days, when that heat will help (a little) with drying, but will also reduce the night sky cooling on cold nights when moisture might accumulate. I'd be hard pressed to estimate which of those factors matters more.

    Situations like this make me wonder about the feasibility of making a fully sealed cavity with vapor barriers on both sides, and with replaceable desiccant cartridges accessible from ports in the drywall. Wireless hygrometers would monitor the situation and warn before you need to replace the desiccant cartridge or if there was a roof leak. Usually when I wonder that, I conclude that that system would work great for the first two years, while there were in fact no roof leaks, and then it would be forgotten by the homeowners, and be out of service by the time it was needed. But it's a fun thought experiment.

  2. User avater
    Ryan Lewis - Zone 4A | | #2

    I’m flattered to be written about, I don’t think it’s clear from the original post, which is my fault— this wasn’t a “mistake” in so much as that it just simply wasn’t possible to meet code on the exterior without replacing the entire roof on the house — a proposition that was out of the question for our budget. However, no matter the context, I found this conversation about what to do enlightening.

  3. Keith H | | #3

    It's been discussed here before but why do we assume the presence of moisture (in vapor form) against the roof deck? Is it the presumption that the air barrier or is installation is inherently fallible? The presumption that there are numerous penetrations? Ignoring the legitimate concern of moisture content of new building materials, what's the source of the moisture? Why not address the source?

    1. Malcolm Taylor | | #4


      Blower door tests on houses air-sealed with a great amount of diligence by experienced builders, still show substantial leakage - and that is when the building is new, un-modified, and also doesn't take into account other methods of vapour-transfer like diffusion, transpiration, or bulk-water intrusion. If we could entirely eliminate the movement of moisture thr0ugh assemblies, building would be a lot easier.

    2. User avater
      Dana Dorsett | | #5

      >...why do we assume the presence of moisture (in vapor form) against the roof deck?

      There is water vapor in ALL air, and air barriers aren't vapor impermeable (nor should there be). When the roof deck temperature falls below the dew point of the entrained air in the cavity insulation it draws moisture from the cavity air into the roof deck. Even with a perfectly air sealed vapor retarder in place to control the rate of transfer, some amount of moisture will diffuse through the vapor retarder to the colder parts of the assembly.

      Making it all very vapor tight as well as air tight has risks too, making the assembly less resilient by slowing all drying paths to a glacier's pace crawl.

      In cool/cold climates the wintertime indoor dew point average are usually 35F or higher, and the roof deck temperatures average well below that for weeks/months on end. So in an unvented roof assembly it's good to slow down that moisture diffusion to limit the peak moisture, but still have it sufficiently vapor open to dry toward the interior. Most roofing materials are VERY low permeance, and nothing drys to the exterior through rain / snow /dew wetted roofing.

Log in or create an account to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |