Hello Pros.

I am seriously considering going with unvented roof assemblies for a couple of houses that may not offer viable\thorough attic access to achieve a proper air barrier and thermal boundary between conditioned spaces below the attic. Disclaimer: I am a lean and very flexible guy, so if I can't get to the tough spots then I have to assume such spaces are inaccessible.

In the first project, the attic space is full of ductwork, and the eaves are virtually inaccessible. The existing attic\roof ventilation is poor and it is unlikely that proper ventilation can be achieved at the soffits\eaves.

The thought is to insulate above the existing roof decking. The house is in Northern Virginia. I was thinking about installing a Grace membrane on top of the existing roof decking, then adding 4" of rigid polyiso, followed by a sleeper layer (Lstiburek recommends using 2X4s on edge), a 1/2" layer of plywood (I want to go with 1/2" AdvanTech OSB), building paper (I want to use Grace Tri-Flex), and finally the new roof shingles.

Here are some of my questions:

1. Lstiburek recommends at least 6" of rigid polyiso but my calculations tell me that 4" will be adequate for my climate. Your thoughts on this?

2. Is the 3.5" air space that the 2X4s will produce necessary? I was hoping that a 1" air space would be adequate. Making up for the aesthetics of the rigid foam and the 2X4s may make for a nightmare, or at least a ton of cost.

3. Some folks recommend both rafter cavity and above-roof insulation. I agree with this in principle but practical limitations appear to be slapping me right in the face, as complete rafter cavity insulation will be tough (if not impossible) in this situation. Other folks are just fine with exposed rafters with all the insulation above the roof sheathing. Your thoughts?

4. Should I just tell the homeowners that their best bet is to completely tear off their roofs so the attic spaces are fully accessible? And at this point proper sealing and insulating should be completely feasible, with proper soffit and ridge ventilation also feasible. I just finished one of these types of projects and the results were awesome. At the end of the day, I really was intrigued by unvented roof assemblies as a few other projects I am working on involve cathedral ceilings that will be a bear to properly seal and insulate.

Thanks in advance Pros!


Views: 747

Reply to This

Replies to This Discussion

1. Well if you go with #3 - you had better be running 6 or more to get the proper R-Value, as for that the dew point calculations are bare minimums, I always go higher to take care of any possible imperfections - don't forget two layers & offset

2. no a pure hot roof does not require any gap, but yes a 1" or better yet a 1.5" gap will work - the reason I say 1.5 is to place a 2x4 on side which gives you plenty of surface for sheathing & to make sure your timber screws have plenty of meat (especially if the trusses or rafters are not fully straight)

3. Prefer to keep all above but not if you are going 4" or other issues come up - you can always staple netting to what you can reach then run it down to the ceiling & fill areas you cant

4. do what you are comfortable with & can truly deliver - when there is a will & enough knowledge, almost anything is possible


Great stuff. Thanks Sean!

The IRC (starting in '09?) requires that any space bigger than 30 ft2 has to have access, so you at least may have to install a hatch, unless your inspector doesn't know/care/see/understand/agree with.

Hello, Patrick,

There are a few important considerations that I think people often miss when deciding between or comparing establishing the thermal enclosure at the attic floor or in the plane of the roof.   First, to really do the same level of airflow control at the attic floor as one can readily achieve at the roof plane and gable walls takes a lot more work than is typically included in an attic air sealing scope.  In other words, if you did everything you needed to do in an existing attic, the cost differences between the two approaches would get much closer.  Second, the end result is just not the same thing.  In the case of a not-so-steep roofs (such as in a typical ranch style) and a low attic height, maybe all you gain with an unvented attic is a reasonable place to store candles and a cavity where it is thermodynamically acceptable to run ductwork.  But if there is any kind of useable space in the attic...consider how much it would cost to enclose that much usable space next to the house (instead of on top of it and under the existing roof)?

In terms of performance, we did a study with a very experienced weatherization group in Chicago (CEDA) that compared implementation of attic flat and roof slope treaments in similar style houses.  One thing we found is that the roof slope treatment seemed to have an edge in terms of air leakage control despite the increase in surface area.  And this was a roof slope treatment involving rigid board foam applied between rafters from the inside (we couldn't give much more of a handicap than that!).  You can read more about this study in a report titled "Attic or Roof? An Evaluation of Two Advanced Weatherization Packages" it is available here: http://apps1.eere.energy.gov/buildings/publications/pdfs/building_a...

To your questions:

1) how much insulation on top?  It depends mostly on what you are trying to achieve.  Code minimum?  High performance?  Zero Net Energy?   Maybe we focus to much on energy as the decision parameter.  If durability were a more important decision parameter (and I would argue it should be) then we might ask, what are the chances that somebody will come along and insulate the full depth of the rafters with a fibrous insulation?  In that case, how much insulation would you need on top of the roof sheathing to keep the temperature of the moisture-sensitive condensing surface (existing roof sheathing) above the interior air dew point  (see http://www.buildingscience.com/documents/digests/bsd-controlling-co... ).

2) how much vent space?  You need a 2" air space to manage the risk of ice dams if you're in a location with a ground snow load above 50 psf.  I'm in Eastern Massachusetts and I found that R25 on the roof deck (and nothing below yet) with a 1-1/2" air space on a hipped roof (therefore not really vented from soffit to ridge) was enough to hold 3' of snow on the roof without so much as a hint of an icicle (the other side of the house with the vented attic roof still in place had monstrous ice dams).  I like a back-vented cladding.  If you don't have a high snow load, then you could probably get away with less than 2".   Consider which is going to be easier for the guys/gals on the roof: attaching 4x8 sheathing to the rafters with long screws or attaching 2x purlins to the rafters with long screws and then installing roof sheathing as usual with a nail gun.   I can't answer that one.

3) Over deck + cavity or over deck only.  I think you've already answered part of this: it's challenging to get a whole lot of insulation (e.g. more than 6") above the deck without things looking weird and getting complicated (unless you're Joe, of course).   Think about condensation control and building in the flexibility to add more insulation without tearing the building apart.  In other words, and to paraphrase Brian Butler of Boston Green Building, don't inoculate the roof against further improvement.

4) Tear off the roof?  If you tear off the existing roof, then you should add another floor so the homeowners get something out of the project!  Again, I think we have the tendency to be too single-minded about energy sometimes and we fail to connect with the broader values of  people who own these buildings and pay for this work.  And if you're building a new floor and new roof, then, by all means consider a vented attic. 

If you do an unvented attic (notice we are being careful to distinguish between unvented attic and unvented roof here), then I would strongly recommend that you use a "chainsaw" transition to the wall if the structural configuration allows it.  There is a schematic detail of this on p. 88 of the Mass Save Deep Energy Retrofit Builder Guide available here: http://www.buildingscience.com/documents/guides-and-manuals/gm-mass... and here: https://www1.nationalgridus.com/DeepEnergyRetrofit-MA-RES  

Good luck with this project and your future endeavors!


Thank you very much Ken. You have included a lot of good food for thought.

The idea of adding a story while tearing off the roof is worth looking into.

I remember the early days of my career in home performance (not too long ago) when things were relatively simple (e.g. seal and insulate attic floors until you're blue in the face). You make a name for yourself as a provider of real solutions, and before you know it you are talking with the pros about big-time projects. I think I will put some of these projects on hold for a few months as I take the time to research things much further.




Home Energy Pros

Home Energy Pros was founded by the developers of Home Energy Saver Pro (sponsored by the U.S. Department of Energy,) and brought to you in partnership with Home Energy magazine.

Latest Activity

Hal Skinner replied to Hal Skinner's discussion An example of what an RCC can do with no R- insulation in the walls.
"Mornin Brett. 90 to 85, yes a slight variance there.  Same ASTM test, different scientists,…"
2 hours ago
Bret Curry replied to Hal Skinner's discussion An example of what an RCC can do with no R- insulation in the walls.
"Great information Hal. It appears the thermal emissivity in your test was .85 rather than .90. Are…"
3 hours ago
Juan Roca commented on Dale Stephens's blog post LED Lighting - Garage Door opener interference
"Hi,There is a real problem with remote controls and LED lights. However, there is a solution in the…"
10 hours ago
Profile IconJuan Roca and Elizabeth Coe joined Home Energy Pros
20 hours ago
Casey Gesell posted videos
20 hours ago
Gerald Shechter posted videos
20 hours ago
Hal Skinner replied to Hal Skinner's discussion An example of what an RCC can do with no R- insulation in the walls.
"Bret,  here are the 2 reports I referred…"
22 hours ago
Craig Foley shared George Kopf's discussion on Twitter
23 hours ago

© 2014   Created by Lawrence Berkeley National Laboratory.

Badges  |  Report an Issue  |  Terms of Service