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

Cellulose insulation for Timberframe roof, will it settle?

carsonb | Posted in Green Building Techniques on

I’ve been looking over the myriad options presented for insulating a 12/12 pitch roof over a timberframe, and am gravitating towards an assembly like the one presented here:  So you fur up the ridge beam by several inches, T&G at bottom, taped drywall for air barrier, cellulose, ijoists, zip sheathing that ties into the wall zip sheathing, then 2x4s for ventilation channel, then sheathing for comp shingles.

My biggest concern is that I’ve been told that cellulose settles.  I’ve also been told that they make versions of it that inhibit settling.  My question is, how much of a concern is this?  Is this something I could potentially DIY install like insulation in an attic?  How much of an issue would small air gaps be when I have a foot of cellulose and a ventilation channel above the sheathing?  I’ve heard air gaps can cause condensation, but why isn’t that an issue with an attic and the giant air gap there?

cathedral batts and traditional rafters have bridging, rigid foam between rafters had the same issue and seemed liable to shrinking foam causing leaks, 6″+ of foam on the exterior worried my contractor given the 12/12 pitch, ijoists with vent channels built into the flange seemed more labor intensive than just sheathing over the whole thing and laying 2x4s.

thanks for any input!

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
    Michael Maines | | #1

    Carson, good quality, modern cellulose installed at the proper density will not settle--it's installed at twice the density as loose-blown. It is fiberized, which sort of fluffs up the particles, rather than just chipping them as they used to do.

    I have seen a lot of cellulose installations that were not dense enough, because installing at the proper density (~3.5-4.0 lbs/ft³) can make it difficult to install drywall without adding additional layers to the assembly. Cellulose can also settle if it is repeatedly saturated, which good design and installation should avoid.

    It takes practice, care and relatively heavy-duty equipment to install cellulose at the proper density for dense-packed installations. It can be a DIY project but it's better to find a pro and verify with others that they know what they're doing.

    Air gaps can be a condensation problem, but if you have a proper ventilation channel they are not a concern. Whether vent channels or open attics, air flow removes moisture.

    1. carsonb | | #4

      thanks Michael. I'm in a heating dominated climate (zone5b), so if I follow this correctly the condensation would occur from moisture in the interior making its way through the drywall layer, through 14" of cellulose, and then condense on the bottom of the top sheathing as it hits a cold surface? My confusion here is how the vent channel above the top sheathing dries out the dew accumulating on the underside of the sheathing. I guess the sheathing is vapor permeable so if it gets wet it will dry outwards, but wouldn't it also get the cellulose wet that's touching it or drip onto it if there are gaps?

      1. Expert Member
        Michael Maines | | #11

        Carson, correct, and when moisture makes its way to the sheathing, it will continue through to the vent space, as long as the sheathing is somewhat vapor permeable. To further minimize the risk of moisture getting into the cavity, I prefer to include a variable permeance membrane at the interior, such as Siga Majrex or Pro Clima Intello. But some say that's overkill.

      2. Expert Member
        Dana Dorsett | | #13

        >"...if I follow this correctly the condensation would occur from moisture in the interior making its way through the drywall layer, through 14" of cellulose, and then condense on the bottom of the top sheathing as it hits a cold surface?"

        The terms "condensation" and "condense" aren't exactly correct here. What happens is more accurately called "adsorption" and "adsorb". There won't be any liquid water happening, but the moisture is taken up into the microstructures of the wood fibers as adsorb, a one molecule thick layer of H20.

        >"My confusion here is how the vent channel above the top sheathing dries out the dew accumulating on the underside of the sheathing."

        It doesn't, at least not rapidly enough for your climate zone. A layer of dry OSB (such as ZIP) is a Class-II vapor retarder- less than 1 perm. It would be classified as a "vapour barrier" under Canadian code. Even when the moisture content is high enough to grow mold in the OSB it's still well under 10 perms.

        Venting UNDER the roof deck (per code) puts a layer of dry outdoor air between the insulation and roof deck, and cellulose is VERY vapor permeable. A thin layer of exterior side air barrier between the cellulose and vent channel is fine as long as the conditioned space side of the assembly is low permeance (< 1 perm). The web of an I-joist would wick and distribute enough moisture from the insulation side into the vent side, even if the air barrier/chute is vapor impermeable.

        Regarding settling, open blown cellulose is typically 1.2-1.4 lbs per cubic foot, and will settle 8% over the first decade. While it might take 3lbs+ density to keep wall assemblies from settling, for only 14" of depth even 2lbs is plenty. Seasonal moisture cycling is a major contributor to the creepage that causes settling- another good reason to keep the interior side reasonably vapor-tight, while allowing the cellulose to dry into a vent channel. Cellulose can store quite a bit of seasonal moisture as adsorb without loss of function, but the less wintertime moisture it takes on, the less prone it would be to settling.

        1. carsonb | | #17

          Dana, you are saying that the zip sheathing would not allow drying to the exterior because its perm rating is too low? I'm really confused then, as that is what I seem to keep reading is a great advantage of using tyvek/zip WRBs. So you are suggesting that a top vented roof would not be the best route and instead I should use those thin foam vent baffles to vent under the sheathing? I keep reading that air sealing such an assembly on 6 sides is really important, could this be done with those flimsy stapled baffles? You would suggest that instead of vapor permeable drywall on the bottom (interior side) I should use something like taped sheathing or foam? I was under the impression that air sealing all 6 sides and leaving the option to dry to the interior and exterior would be the most robust approach since most moisture comes through air leaks not vapor diffusion. This also seems to be what Michael is suggesting (vapor permeable bottom layer).

  2. Expert Member


    There are two types of cellulose installations. Loose-fill, which is blown onto horizontal surfaces like attic floors, and dense-pack, which is compressed to a density that precludes settling. Dense-pack done properly doesn't settle in walls, and would not settle in a sloped roof cavity.

    1. carsonb | | #6

      thanks Malcolm, sounds like DIY is a probably a bad idea then. I was hoping it would be roughly equivalent to an attic and perhaps the sheer weight of it would dense pack itself.

      1. Expert Member
        Michael Maines | | #12

        Carson, dense-packed cellulose installed properly--so it won't settle--is very firm, like a firm mattress. If you were to blow an attic very deep--maybe 48" or so--the bottom portion might get to a dense-packed density, but in a typical rafter bay using a rented machine you will not get it dense enough to prevent settling.

        My friend Jon is the best cellulose installer I know, and he wrote a good article about it:

  3. Expert Member
    Akos | | #3

    Around me dense pack is pretty expensive. For something like 12/12 you can't loose fill as it will settle.

    You are better of getting batts for metal studs. These are wider and fit snugly between I joists. Two layers of 2x6 batts with a foam vent baffle make for a pretty low cost roof. This also avoids the issue with the weight of dense pack and bowing your ceiling.

    There is no need for two layers of roof sheathing, if you need more insulation, use larger TJIs.

    For best results your ridge beam should be bellow your I joists with the air barrier (angled 1/2 plywood wings with taped seam) across the top of it before the TJIs are installed. Your drywall now butts up against these plywood wings and is sealed with a good quality tape. You can then clad the beam to make a nice accent piece.

    1. carsonb | | #5

      Thanks Akos, I liked this idea primarily to completely reduce thermal bridging and ice damming. I had considered your earlier suggestion of fastening foam to the top flanges of ijoists for vent channels but my concerns were with ice damns, the foam creating a condensing surface (wouldn't I need the foam to be about 4" thick to avoid this? Actually, how do the manufactured flimsy foam vent baffles not have this issue either?), and labor costs of fitting foam between the flanges was more than just sheathing over the whole thing. The other consideration I had was air sealing. Zip sheathing over the top seems to my non-expert mind to be the most straightforward way to air seal the top of the assembly, which can also be tied into the zip sheathing on the walls. The foam vent baffles would be just stapled to the sheathing right? How are the baffles installed if they can't be accessed from below?

      1. carsonb | | #7

        Perhaps I'm trying to be too creative here, but another option to avoid any bridging and still use cheaper batts may be to just fasten a few inches of foam to the bottom of the ijoists. that sounds very easy and cheap.

      2. Expert Member
        Akos | | #9

        The vent baffle should be minimum thickness somewhat permeable. You want something like OSB, fiberboard or unfaced rigid insulation. This way any moisture that makes it in there can move into the vent channel.

        If you don't need any extra R value it can also be cardboard or house wrap stapled up. Either way, it should be less cost than the double roof. The baffles are installed from bellow after the rafters go up but before the insulation goes in. The simplest is to cut a roll of housewrap on a miter saw to a bit under the space between your TJI webs, unroll and staple as you go along.

        It is pretty simple to get air barrier continuity to the celing over top plate, see my reply here:

        You can use foam as the warm side air barrier instead of drywall. Foil faced rigid is pretty easy to air seal and you can nail T&G directly over up to 1" of foam.

        The top vented roof is good for snow country, so if you have very heavy snow in your area, it is worth it.

        1. carsonb | | #15

          it sounds like I have more reading to do. If I were to use OSB for the vent baffle though, wouldn't that be the same thing as the "double roof"? It sounds like more labor to fasten it on the underside of the flanges and would be harder to air seal? We get about 3 feet of annual snowfall here on average.

          1. Expert Member
            Dana Dorsett | | #19

            If you're going to dense pack you can't use thin foam for the baffle- even 1" foam is too flexible, and will buckle.

            While half-inch OSB would be sufficiently structural as a it's also a class-II vapor retarder when dry. It'll work, as long as the ceiling side of the assembly is reasonably vapor retardent. Half inch MDF or asphalted fiberboard sheathing will bow over time unless supported between the rafters by spacer to the roof deck (1 foot on center between support points, either I-joist flanges or spacer.) The advantages of MDF or asphalted fiberboard is the high vapor permeance (15+ perms), and a high tolerance for moisture, offering superior drying rates toward the vent channel.

          2. carsonb | | #24

            thanks Dana. From several suggestions here it sounds like blown insulation might not be the best thing to try on a 12/12 pitch roof (although it seems like the cheapest and lowest embodied carbon). If I do go with batts, do you see any downside in using the housewrap for vent baffle as Akos suggested? It doesn't seem like it would be terribly good at air sealing very well, but no one in this discussion so far has advocated for the importance of air sealing the top of the assembly.

        2. carsonb | | #18

          I'm confused how important the 6 sided air sealing is then. It seems like sealing 3" or so of foam to the flanges of the ijoists would be tougher to seal affectively than zip over it. Foaming the joints would potentially lead to cracks once that foam shrinks correct?

          1. Expert Member
            Akos | | #20

            You only make the vent baffles out of fancier material if you need extra R value. Around me code is R31 for cathedral ceilings so R30 batts + fiberboard (or R28 + R3 foam) is a good combo.

            Rest of the time you make the baffles out of the cheapest material. Ripping a stack of 1/4" osb and stapling it up to underside of TJIs is a quick. Building the 2nd roof on top is about the same work as building the original roof. They are not comparable.

            Another low cost option would be to build a small attic at the peak of the roof while keeping the rest of the ceiling cathedraled. You can then fill everything with loose fill insulation but add a couple of inches extra insulation to the mini attic.

            Any settling even in the cathedral section, will just reduce the insulation thickness in the attic area only.

            With high density batts (mineral wool or high density fiberglass), I don't think the six sided air sealing buys you much if anything. For my own home I never bothered with baffles past soffits.

            Important part is a very good warm side air barrier, the rest are bonus if easy. If this costs you a lot of labour, you are better spending that money on increasing the insulation thickness a bit to compensate for any wind washing effects.

          2. carsonb | | #28

            Martin writes that air leaks in the baffle will cause negative pressure due to the stack effect, thus forcing interior air from the interior through the bottom of the assembly. How would you air seal the housewrap used for a ventilation channel adequately? Tape it to the tjis?

          3. Expert Member
            Akos | | #29

            Air sealing the vent channel is generally too fussy of a work to be worth it.

            What you do need is your ceiling sealed well. This should be as solid as possible with preferably two layers, the first layer could be your drywall and the 2nd layer a well detailed poly vapor barrier (could also be one of the fancier membranes).

            Use only sealed electrical boxes for lights and stay away from pot lights in the ceiling.

            If you take the above precautions you will have a robust, high performance assembly.

    2. carsonb | | #23

      Akos, do you know of an article/diagram of this detail?

      "For best results your ridge beam should be bellow your I joists with the air barrier (angled 1/2 plywood wings with taped seam) across the top of it before the TJIs are installed. Your drywall now butts up against these plywood wings and is sealed with a good quality tape. You can then clad the beam to make a nice accent piece."

      1. Expert Member
        Akos | | #25

        See attached sketch.

        1. carsonb | | #26

          I feel like I should be paying you a consulting fee:)

  4. Paul Eldrenkamp | | #8

    "Settle" may not be the right word, but even dense-packed cellulose in a roof assembly can compress over time, at least on north-facing slopes. It can compress as much as an inch or two over 8-10" of installed thickness. It does not require saturation. I think it has to do with high relative humidity levels that result from the extreme cold temperatures at the sheathing caused by night sky radiation. Kohta Ueno has spoken about this at Building Energy conferences, and I've seen it myself on a couple of projects.

    1. carsonb | | #10

      Do I trust the insulation contractors that new formulations and appropriate dense packing solve this issue? When it comes to issues with my roof, I tend to want to go for the safest approach possible.

      1. Expert Member
        MALCOLM TAYLOR | | #14


        If you aren't confident i'd find another solution. I wouldn't trust the cellulose installers I use for loose-fill to dense-pack a roof like that either.

        1. carsonb | | #16

          Malcolm, so you would use batts as the safest, most cost effective option?

          1. Expert Member
            MALCOLM TAYLOR | | #21


            I think I'd follow Ako's advice.

    2. Expert Member
      Michael Maines | | #22

      Paul, if I recall correctly, Kohta was showing images of an unvented roof assembly with compressed cellulose, at the seminar I saw anyway. I thought he said that if the roof was vented it would probably not have happened. I supposed it would depend to some degree on how easily moisture can move through whatever baffle is between the cellulose and dry air. Have you seen compressed cellulose in properly vented roofs?

  5. carsonb | | #27

    Martin writes about air sealing ventilation baffles: "Paying attention to airtightness when putting together the components of your cathedral ceiling assembly is the most important thing you can do to limit the chance of moisture accumulation and rot. This applies to the ventilation baffles as well as the drywall ceiling.".

    this would seem to be cautioning against using leaky baffles such as stapled tyvek or any of the manufactured baffles. How would you adequately air seal them?

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |