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

Low-Slope Unvented Roof for Passive House

qbrt | Posted in General Questions on

I’m in the early stages of planning for a passive house in NYC, so no architect or builder yet. I’m trying to iron out a lot of details before I hire someone so I can show them what I have to see if they’re comfortable with the kind of thing I’m looking for.

Here are my design goals:
* Passive house compliant (~R-67)
* Flat roof
* Solar panels on top
* No spray foam (for maintenance, health, and environmental reasons)
* No cellulose (for fire safety reasons)

I’ve been reading everything I can find, but roof construction is very confusing for a novice! I just finished working through NYC 2022 building code (based on IBC 2015) and doing all of the load calculations. To get 25′ spans, I’ll need to use 2x12s. While this will probably be expensive, this gives nice deep rafter bays to fill with insulation. Given my constraints above, fiberglass and Rockwool seem to be my only options to fill the bays.

That leads me to a design something like:
Plywood/OSB deck
5.5″ XPS (R-22)
11″ Rockwool (R-42)

Is this a reasonable design? I would really appreciate a sanity check or ideas for improvement! Also, where is the right place for an air barrier and WRB barrier in this?

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. Expert Member
    1. qbrt | | #2

      Thanks, this seems to roughly validate my approach -- the articles and comments took me down a chain of articles that I think answered my questions :)

      In my CZ, I need 31% of my insulation to come from the foam, and I'm at 32%. That's maybe cutting it a little too close for comfort, so maybe I go down to 9.5" of Rockwood for R-38 and 6.5" of R-4.5 rigid foam for the rest, which would give me 44%. says the WRB can either go underneath or on top of the foam. Since I'll probably go with an integrated air/water barrier (like zip system), I think that answers all of my questions. Thanks!

  2. Expert Member
    Akos | | #3


    Dense packed borate treated cellulose simply won't burn in any house fire. Similar to mass timber structures, there is too much mass, it might char but that is about it. By that time there is probably nothing left of the house.

    If you are going for actual passive house cert, than by all means build an R67 roof.

    If you simply want an efficient structure than R67 built with any foam makes zero environmental or financial sense. Go for code min based on U factor (most likely an R38) assembly. When it comes to heat loss, a roof is pretty much another surface, it makes no sense to have an R67 roof unless you also have an R67 wall. You don't need to take my word for it, I'm just a random guy on the internet, have somebody run a calculation on your yearly energy use for the different options.

    When dealing with hybrid assemblies with rigid insulation, the most efficient house has roof and walls about the same R value. You do want slightly higher R value for the roof as you do get more radiative loss at night time but not 2x walls.

    As for your assembly, since it is a roof there is no WRB. Water barrier is the actual roofing. You do need a couple more control layers bellow the roof membrane.

    From the inside out.

    You need a warm side class III vapor retarder and air barrier. The simplest is painted drywall.

    Above that goes the batts or dense pack insulation between the rafters.

    Your roof deck should be your main air barrier. Plywood or OSB with taped seams is the simplest. Make sure to figure out how to tie this to your wall to ensure air barrier continuity.

    Rigid insulation above this and your membrane roof either directly onto the rigid or over a cover board.

    The last thing you want to use in a green build is regular XPS. It is blown with a high GWP blowing agent which diffuses very quickly so the long term R value is about the same as EPS. You want the newer low GWP XPS, EPS, GPS or polyiso.

    Most membrane roofs are installed over polyiso, depending on the thickness it is around R5.9/inch. Even for your R67 roof you only need 3.75" or so of polyiso.


    1. qbrt | | #4

      Thanks, I really appreciate your detailed response! If a good roof should be insulated similarly to a wall, why do so many passive house programs recommend such a high R-value?

      With cellulose, my understanding is fires can start in them relatively easily, but they spread extremely slowly. In practice, this means cellulose insulation can smolder for hours.

      1. Expert Member
        Akos | | #5

        The high R value roofs are recommended when you are using cheap insulation like blown in fiberglass or cellulose. With these the incremental cost of more insulation is essentially noise compared to the house budget, even R100 is easy.

        Once you add in rigid insulation or spray foam into the equation, the incremental R value costs goes through the roof. With high R value roofs, in terms of simple ROI on energy saved you are looking at centuries.

        Not the most scientific test, but I've put a torch to standard borate treated cellulose. It chars but that is about it. Besides the low cost and reduced carbon footprint a nice side benefit of a couple tons of cellulose in your walls and ceiling is that it makes for an exceptionally quiet house.

      2. Expert Member
        BILL WICHERS | | #7

        XPS and EPS, when they catch fire, create melting blobs of flaming goo that spreads fire. Polyiso chars, so it's safer. Fiberglass and mineral wool are essentially non-combustible, so they're safest in terms of fire resistance, with mineral wool being superior to fiberglass in that regard.

        Aside from that though, I agree with Akos here -- cellulose has been used for many decades, and it's not known as a fire risk. If your main concern is fire risk, I would pay more attention to fire blocking and air sealing, which will help limit the spread of fire and smoke migration. I would also use 5/8" drywall throughout to slow the spread of fire. You could also put spring hinges on all the doors to keep them closed, which also helps to limit fire spread. Running all your electrical wiring in conduit will also help provide some additional protection.

        What I'm getting at is that cellulose is not a big contributor to fire risk in a home. There are other things that are more of an issue. If you really want to go all out, you could even put in a sprinkler system. Using 5/8" drywall is a cheap way to really improve things too, with the added benefits of much more solid feeling walls, and less sound transmission between rooms.


    2. Expert Member


      I've seen this comment thrown around before - "When it comes to heat loss, a roof is pretty much another surface, it makes no sense to have an R67 roof unless you also have an R67 wall. " But I'm not sure I quite understand the physics of what could make that true.

      The exitance from the sun is mostly the irradiance on the roof, and not the walls. That's where the energy is coming from, trying to make its way into your house, through the roof. Perhaps its different in cooler climates, or more equal I should say, but in the south it's very apparent where they energy is coming from, and it only makes sense to slow its transfer down as much as possible, IE more insulation.

      Not calling you out, I'm just trying to figure out why I keep seeing this thrown around.

      1. Expert Member
        Akos | | #8

        In colder climate, your heat loss is determined by the temperature delta between indoors and outdoor, area of the surface and R value of the assembly.

        About the only difference between a roof and a wall is that a roof will see a bit more loss during a clear night due to radiative loss and gain a bit during day as the sun will warm the roof surface. Overall these two effects seem to cancel out in most climates.

        When it comes to cooling load, it is a different story. In that case most of your heat gain comes from glazing, air leaks and sun beating down on the roof. Even on a sweltering hot day, the temperature difference between indoors and outdoors is not that much this is why wall R value doesn't matter too much.

        For example, around me my winter heating design delta T is 68F. The summer cooling delta T is only 9F.

        Overall, in a heating dominated climate, a roof is just another surface no different from a wall. 1000 sqft of R30 roof will have about the same heat loss as 1000 sqft of R30 walls.

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |