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
Q&A Spotlight

How to Attach a Thick Layer of Exterior Insulation

A reader worried that extra-long screws will sag over time looks for an inexpensive and practical solution

Burke Stoller proposes using sections of PVC pipe as standoffs when attaching a 6-inch later of mineral wool insulation to the outside of his walls. He hopes the pipe sections will help prevent the fasteners from sagging over time.
Image Credit: Image #1: Burke Stoller

Adding a layer of insulation to the outside of a house, over the wall sheathing, makes all kinds of sense from an energy perspective. But the thicker the layer, the more challenging becomes the actual means of attaching it to the building.

In a post in the Q&A forum at Green Building Advisor, Burke Stoller shares some of his concerns, as well as a proposed solution. Stoller is working out the details for a 6-inch-thick layer of Roxul ComfortBoard mineral wool, consisting of two layers of 3-inch-thick panels, each 2 feet by 4 feet.

“My concern is that with such a substantial thickness of ComfortBoard, there is a potential for faster ‘sag’ through this, or durability issues during seismic events,” Stoller writes.

“On many of our projects, we have screwed our vertical cedar 1×3 rainscreen material directly through a single, thinner layer of the ComfortBoard directly into the framing,” he continues. “With these thinner layers (2 inches or less), the assembly seems reasonably solid and durable once all the rainscreen strips are attached. It does require some fussing around with sucking screws in and out to keep all of the strips co-planar and flat, but it’s not too bad.”

But a 6-inch layer of insulation poses different challenges. Deflection over the long term may threaten the durability of the attachment.

Stoller’s proposed solution is to insert lengths of 3/4-inch schedule 40 PVC pipe through the insulation and then run long screws through the pipe as he attaches his rainscreen batten.

“What do people think?” he asks. “Would this work, or does it just seem like a lot of work that won’t actually provide any resistance to sag, or deflection? Also, I am wondering if I’d be best served by using 7 7/8-inch-long GRK screws, or 9 3/4-inch GRK screws?…

GBA Prime

This article is only available to GBA Prime Members

Sign up for a free trial and get instant access to this article as well as GBA’s complete library of premium articles and construction details.

Start Free Trial


  1. AsimM | | #1

    Great article!

    I am trying to detail something similar for my exterior walls. I have a question for the experts. My wall system has steel studs with no sheathing and 3" of foam on the outside. I am unable to find long enough wood to metal or metal screws of the length required to attach furring strips to the studs.
    Would I be ok using wood screws? Any suggestions?

  2. GBA Editor
    Martin Holladay | | #2

    Response to Asim Majeed
    My advice is that you consult an engineer. My guess is that an engineer is already involved -- after all, if the wall has no sheathing, you probably need an engineer to approve your wall bracing plan.

  3. AndyKosick | | #3

    Misunderstood Assembly
    I simply have to chime in here because the drawings and suggestions above seem to indicate the misunderstanding I used to have of exterior insulation assemblies. A couple of caveats, I am not an engineer, I've just had this explained to to me by really smart people that literally test it in laboratories and I also cannot speak specifically to mineral wool, only foam plastic.

    The key is understanding that the fasteners are not holding the assembly up, only together. The consistently underestimated strength of these assemblies comes from the friction created between the large surface areas of each layer. (In example, take two small pieces of foam sheathing, lay one on top of the other and slide the top one across the bottom one, now set a concrete block on top of them and try to slide the top one.) The suggestions above such as PVC pipes and putting screws at angles all sound like they might compromise the fasteners ability to hold the layers together tightly and actually make it weaker.

    My confidence in these assemblies also comes from the fact that, despite the concerns of every builder I talk to, I have never seen or heard of one these assemblies failing in the way everyone is so concerned about. If someone has I'd love to hear about it and so would the really smart people that explained this to me.

  4. user-5714885 | | #4

    Similar wall design
    I have almost the exact same design concept for my wall- 2 layers of 3" Roxul board. But I don't think I can use the Heco Topix screw, I need something that can be used with metal studs and metal hat channels (instead of wood furrings).

    I e-mailed Roxul some time ago about using multiple layers of Comfort Board, they were kind enough to give me the following response:

    1. Can I use 3 layers of 3" Comfort board in a wall assembly? Are there any issues with this?
    Yes – 3 layers of 3” can be used. The concerns would be ensuring that the fasteners can support the weight of the cladding. If thick insulation and a heavy cladding is used, I would recommend intermittent blocking to help support the load.

    2. Can this product be used above a roof deck (i.e. sandwiched in between roof deck and standing seam roof)?
    Yes –The product is not tested for this application but it would be doable provided limited load is placed on the COMFORTBOARD. All loads should be designed to be supported by metal deck and supports.

  5. Expert Member

    The calculations of the shear and bending strength of fasteners take into account the ratio of exposed shank, depth of penetration and what material they are sunk into. I think the previous posts have taken into account the forces you are describing.

    I do wonder though if all of these forces, especially the effect of friction to stop the assembly from slumping, can be relied on to remain constant over time? Foam shrinks, and mineral wool is quite hard to compress to a point where it would act in the way you describe. I'd still want to be confident that, without the insulation adding to the strength of the fasteners, they could support the weight of the siding.

  6. Irishjake | | #6

    I was told by my engineer that where I live (zone 6a/Central NH), and we have 80 lbs of snow load, that even three inches of roof insulation won't pass muster. The funny thing is, is like some have alluded to, a VERY well known building scientist from MA lives in a house with LOTS of rigid foam on their roof. When asked about their assembly - the answer was "we are an engineer and architect and we feel fine with the assembly - plus last winter we had 8 feet of snow(!) and no issues"......

    There you have it.....where are these failures. My solution at my house was to support all the wall insulation from either a brick ledge or from the footing up, with flashing details above grade to break the lower insulation from above, for gross water infiltration and ants, etc..

  7. Expert Member
    Dana Dorsett | | #7

    Lstiburek has CLOSED cell the underside of his roof deck.

    "In went some skylights on the back-side (Photograph 1) and the rafters got sprayed full of 2 lb/ft3 density spray polyurethane foam (SPF)"

  8. JC72 | | #8

    @Brad Hardie
    Just an FYI..

    Lstiburek installed 10 inches of foam on the roof (5 layers of 2" polyiso) of his barn.

    Now if you're talking about his house he already has CLOSED-cell spray foam on the underside of his roof sheathing and then added 6" of exterior rigid foam on top of the roof deck (3 layers of 2" polyiso).

    Lstiburek has a good article on installing an air gap between the roof and the foam underneath for areas where there's a lot of snow accumulation.

    Edit: Open cell was incorrectly cited. He had closed cell. Thanks Dana.

  9. Irishjake | | #9

    6", 10", 12"
    Yes, Chris you are correct - Lstiburek's house was 6" on the roof, and their barn was 10". REMOTE Wall systems spec's 12" on the roof, and even shows how to install it in their video series on REMOTE Walls.

  10. user882465 | | #10

    Screw Length and Sagging
    Two comments - and maybe those with greater experience can comment.
    You should look for long screws at a roofing supplier. They are not surprised by wanting to screw through 6-8 inches of insulation.
    My understanding is that most rigid insulation products have sufficient compressive stress so that the screw is not cantilevered. The screw and the insulation combine to form a truss of sorts. The screw is in tension and the rigid insulation works as a compressive member. Thus the attachment point works structurally as a triangulated truss.

  11. charlie_sullivan | | #11

    Compressive strength of insulation for triangulation
    In reply to Bill Burke: The concept of using the compressive strength of the insulation in combination with the tension on the screw to create triangulation only comes into play once it has sagged a smidge, such that the screw is at a bit of a diagonal. Prior to that point, it's only friction and the screw acting as a cantilever beam that is holding the siding up. And even after it's sagged a smidge, the triangle is so skewed that ratio of compressive force on the insulation to vertical support is huge. Mineral wool board has something like half the compressive strength of even low density EPS. And as it gets thicker the triangle is more skewed for a given limit on acceptable sag, while the deformation for a given compressive force goes up. So there are a lot of reasons that what works for moderately thick foam might not be adequate for thicker mineral wool.

  12. ryem | | #12

    double stud vs ext. board insulation
    Working with Ken Ruddy of Fiddlehead Construction in VT we have been grappling with these same issues. We have settled on a double stud with roxul system we have used on the past 3 houses, and are quite satisfied with so far. The layers are as follows.

    A 2x4 wall 24oc, sheet rock inside, taped and sealed sheathing outside. This is the primary framing and service cavity. Gets 3.5" r-15 roxul batts.

    Then EPS ripped into 2x4's are used on edge as spacers to prevent thermal bridging as the 2nd wall gets lagged into the first. Timberlocks are driven 24oc through the exterior framing, eps spacers, and mid wall sheathing into the interior framing. Bottom plates are included. Exterior foundation insulation continues the plane of the cantilevered wall.

    The double framing creates a 7.25" cavity for r-30 roxul batts. That is sheathed and taped with the Zip wall system. A continuous rain screen foundation to roof is created by 3/4" strapping followed by clapboards.

    I might try 3 or 6" of comfort board in place of the Eps sometime. Still use the double studs with 3.5" batts between. I would increase the length and frequency of screws if under any doubt. I am also a fan of the staggered angled pattern for locking things down.

    The air barrier is located 1/3 from the inside, with enough insulation on the exterior to prevent condensation and no poly to facilitate drying in both directions. Paint serves as the interior vapor barrier.

  13. Expert Member
    KOHTA UENO | | #13

    Response to Asim Majeed (Steel Stud Screws)

    I am trying to detail something similar for my exterior walls. I have a question for the experts. My wall system has steel studs with no sheathing and 3" of foam on the outside. I am unable to find long enough wood to metal or metal screws of the length required to attach furring strips to the studs. Would I be ok using wood screws? Any suggestions?

    There is a commercially-available system that is very similar to what you are describing--a cladding attachment system that directly attaches to steel studs through layers of foam:


    The screws that are used in their system are called out as: "STAINLESS STEEL TEK SCREW w/1,000 HR SALT SPRAY COATING AND ThermaStop™ THERMAL ISOLATION, BY KWS." I doubt that Knight Wall would be interested in selling the screws instead of their system, but you never know.

    The closest screw I can think of offhand that might meet your requirements (long screw, drill tip for penetrating steel) would be screws used in commercial roofing:

    XHD (Extra Heavy Duty) #15 Roofing Fastener

    RetroDriller Fastener

    Apparently, Fastenmaster Headloks (the go-to screw for furring over insulation) are available as drill points, but only 6" and longer:


    For reference, NYSERDA's study on cladding attachment studied steel stud systems (see page 18, for instance):

    Fastening Systems for Continuous Insulation

    Of course, I agree with Martin that you need to be sure that you have addressed shear/racking in the wall before omitting structural sheathing. BSC's guidance on the topic is available here:

    GM-0902: IRC FAQ: Wall Bracing Requirements for Insulating Sheathing

  14. Expert Member
    MALCOLM TAYLOR | | #14

    How much additional labour do the foam 2"x4"s represent? I like the idea. Here in the PNW I'd just be using 2", but don't have a handle on how cumbersome it is to do.

  15. StollerB | | #15

    Foam and Mineral Wool are Incomparable
    Andy- (post 3). I appreciate your clarification of the forces at play in exterior rigid foam applications, and understand how well they must work. However, as I was installing 1x3 cedar rain screen at work today overtop of 2" of Roxul Comfortboard, I was thinking about your comments. They simply don't apply when it comes to exterior high density mineral wool. Their "high density" is nothing at all like the density of XPS, or even Type II EPS. Time and again, to maintain a flat, co-planar wall, I would sink the screw into the rain screen overdrive it sightly to get the head to set flush with the face of the cedar. By then, however, the 1x3 would have crushed the COMFORTBOARD down about 5/16" beyond it's normal 2" thickness, and would be substantially further inboard of all the other rain screen strips on the wall. So, I would then have to reverse my drill and back the screw out until the Roxul "puffed" back up to it's 2". To keep that Roxul at it's 2" depth, the screws are just barely applying pressure. It is NOTHING like the amount of torque one is able to apply to a rain screen strip over a piece of rigid foam with almost zero noticeable compression. Until one has actually stood there with drill and rain screen in hand and installed it over a piece of COMFORTBOARD, it is hard to impress exactly how uninspiring that physical connection is. And that is only over 2"! That is why I am so concerned about what 6" of exterior mineral wool would act like both over time, and in a seismic event.

    I have absolutely zero concerns about the ability of rain screen over rigid foam to endure sagging forces. Mineral wool, however, is a completely different scenario. One would absolutely need some kind of supplementary support rather than relying on the surface tension between the strap and the surface of the insulation, or the resistance to compression of the mineral wool (which as I can tell on site is not as effective in a real world setting as the numbers may seem to indicate). And all of those measures to reinforce the connection of the rain screen add significant material cost, complication and labour to the exterior insulation process, which is what I was butting my head up against as we all worked through the issue in the original Q&A post.

    I came up with a great analogy this afternoon as well, while nailing a piece of perforated metal bug screen onto the 1x3's over the COMFORTBOARD. The amount of bounce experienced on that 1x3 as I hammered away to drive that nail in made me exclaim to my co-worker: "Sheesh! It's like trying to run on a waterbed!" That pretty much sums up the Comfortboard installation procedure for me, ha, ha! I love the stuff, but it's a weird duck to work with.

  16. John Straube | | #16

    The exterior attachment detail has been solved
    A perplexed GBA reader forwarded this article to me. To paraphrase the reader: “Have we not already solved the exterior attachment detail? I have seen people showing numerous photos of projects.”

    Er, yes, in fact, we have.

    We have done laboratory tests, and we have built many projects with a range of exterior insulation.

    It is true we have little experience in the field with 6" Roxul, but we DO have a lot of experience with 2-4" and none of it suggests there is a concern with short- or long-term cladding.

    The comments even suggest that a measly 80 PSF snow load is a problem for insulation rated 280 psf (2 psi) or more!

    See the attached photo (below) where we have taken the concept further by using the furring strips and insulation to support the overhang (and a deck too, if that is interesting).

    And we used standard cheap roofing screws (large-headed screws compress the insulation more and cause more workmanship issues) AND Roxul Cavity Rock DD, which has HALF the density of Comfortboard (which is a bit unyielding, more expensive, harder to handle, and more resource-intensive).

    We spent over 25 years proving to people that you could use foam insulation to support cladding, building literally thousands of houses in all climate zones. There are still people who don’t believe this or choose not to believe. It feels like we are at the start of a 25-year process for stonewool. People don’t want to do the homework of reading the many reports of technical performance and the case studies (see and or invest a few hundred dollars to build an 8'x8' mockup, or even investigate a source of screws other than their local Ace and Home Depot.


  17. GBA Editor
    Martin Holladay | | #17

    Response to John Straube
    Thanks very much for your comments.

    GBA readers who are interested in learning more about Dr. Straube's laboratory tests should read Installing Mineral Wool Insulation Over Exterior Wall Sheathing.

  18. Expert Member
    MALCOLM TAYLOR | | #18

    I think we are talking at cross purposes. I don't think anyone suggests you can't support cladding over foam or mineral wool. The question is what size and depth of fasteners to use for various thicknesses of insulation. There haven't been failures probably because builders have so far been conservative in how they attach the furring. Obviously there is a minimum spacing and size necessary, and if you go below that thing will go wrong. The question being asked here is: for 6 inches of mineral wool, what is that?

  19. StollerB | | #19

    Climbing Wall
    I appreciate your comments John Straube. I'm sure you feel like a broken record sometimes, but that is probably a symptom of being an in insular segment of the engineering world filled with extremely progressive colleagues! I'm sure we all wish we could go down to our local chapter of a BSC franchise and employ such services on our jobs.

    This morning I was installing vertical 1x3 cedar rainscreen @ 16" o/c over 2" thick Roxul COMFORTBOARD, with horizontal strapping for cedar shingles. After finishing, I literally climbed up onto the horizontal strapping so that it was bearing all of my weight and bounced up
    and down. It didn't budge, compress or bounce in the least with a dynamic 180lb load. Horizontal lap siding over verticals would achieve the same type of rigidity.

    As Malcolm indicated, the rigidity of rainscreen over these kinds of thin levels of Roxul was never in question, and your research and the field experience of many has proved this. But the fact that Roxul suggests anything beyond 4" of COMFORTBOARD requires the input of an engineer suggests that at increased thicknesses other dynamics may come into play. That is what the original question was interested in discovering.

    Do you suspect that the increased unsupported screw length in a 6" layer would not in fact cause any issues that don't appear in much thinner layers?

  20. John Straube | | #20

    Response to Malcolm Taylor and Burke Stoller
    The article began with the suggestion that PVC pipes, special screws, sloping screws, or clips be used to solve the problem that “Deflection over the long term may threaten the durability of the attachment.” This was not posed as a solution to the problem of not knowing what fastener pattern or diameter to use.

    A report on Comfortboard is on the Roxul website reports results for 3” ComfortBoard:
    Cladding Attachment and Support

    Table 4 of this report shows that with #10 screws at 16” o.c. vertically and horizontally, that siding (and even 12 psf stucco) will deflect less than 1/100” with 3” of CIS if furring strips are used.

    So, why would someone be concerned with 6”? Lets say deflection goes up by a factor of ten times when we double the insulation thickness. (There is no engineering physics to support that assumption; engineering would normally assume no worse than 4 times and as little as 2 times depending on the mechanism assumed.) So would 1/10” deflection be a problem? I doubt it!

    Numerous reports have been presented over the years that have built confidence and an approximate understanding. We still don’t have a precise predictive method, and likely never will given the variability of the job-site conditions and products used. But because the results always show such small deflections, and such little movement over time, the DOE report states, “Most common residential cladding systems (metal, vinyl, wood, and fiber cement) are lightweight enough (
    If it is a test of thick insulation you want, our lab has tested thicker insulation, lots at 4” some up to 8”. Take a look at the Building America 1314 report, which, in Figure 5, shows our results from testing 8” of mineral fiber (an 8 pcf product, less dense than Comfortboard):

    This showed good stiffness up to at least 30 pounds per fastener. So you could figure out how heavy your cladding and furring is (usually it is in the 3-5 pounds per square foot range) and decide to stay well below 30 pounds.

    In practice, we often find that 15 pounds/fastener does not limit us with regard to normal design, because we need to put the fasteners closer together to avoid failure of the furring strip under negative wind loads.

    Thus, if one uses 16” vertically, and 16” horizontally, you are safe.

    We rarely would use #10 screws, not because they are usually special order when they are 6” or more long, but because they are more likely to twist off when they are that long.

    So say you use 7.5" or 8" long #12 screws for 6" of insulation. That would be quite conservative as the screw is larger, and the insulation thinner than the 8” thickness tested, so the stiffness, strength, and creep resistance will be higher than the tested thickness.

    Of course Roxul will not design your wall system for you. Of course they have to ask you to get an engineer. They can't take responsibility, and neither can I. This is the same as pretty much all building design.

    I can tell you that it works, show you case studies, etc. -- but you have to collect and understand the information and make decisions.

    Given the cost to hire a knowledgeable, or to cost of not knowing, it seems to me that people with this kind of question would simply build a 4x8’ test wall like we did, and hang some weights off it. Cheap, definitive, and helpful. In most cases, that is much cheaper than hiring an engineer, who will likely over-engineer the solution because of a lack of experience. (Can't blame them on that).

  21. GBA Editor
    Martin Holladay | | #21

    Very helpful
    Thanks, John. That's a very helpful response. It should provide reassurance to builders who plan to install 6 inches of mineral wool on the exterior side of their wall sheathing.

    I have edited the article to direct readers to Comment #20.

  22. Expert Member
    MALCOLM TAYLOR | | #22

    Very helpful Indeed
    That's a great common sense approach - and the rules of thumb make a good starting point.

  23. StollerB | | #23

    I really appreciate your summation on the behaviour of much thicker levels of insulation, and the Building America reports was terrific. I am stunned at the effect that relative humidity had on the deflection! I never would have guessed at something like that, but it was both direct, and pronounced. I am also very surprised and pleased by the deflection data surrounding both the 8" and 4" mineral wool product using no special techniques beyond what one would utilize for thinner levels.

    Thank you very much for sharing this information. It gives us much greater confidence and clearer guidelines for exterior insulation/rainscreen/cladding systems using mineral wool, and what kind of fastening protocols are appropriate. What that clear direction, we will easily be able to do field mock-ups along those lines that we can then use to satisfy our own judgement about what is acceptable for our particular situations.

  24. charlie_sullivan | | #24

    I still have questions
    John, thanks for pointing us to the BA-1314 report. That contains real data answering many of the questions that we've been speculating about, and more. It's a tremendous resource.

    However, I'm experiencing a little cognitive dissonance--your posts seem to say that this is a solved problem and there's no need for all this discussion, and it's followed by a chorus of smart people agreeing. But the conclusions of the report are roughly that although these systems seem to work fine, we don't know why. I don't find that very reassuring. I see that you suggest building a 4x8 test wall, and I agree that that could be cheaper and more certain than hiring an engineer, especially if the engineer is not familiar with this stuff. I'd be comfortable going ahead with a wall based on such a test, but the need to do that will deter a lot of builders. So I agree with the conclusions of the report, that more data is needed.

    Some of the specific things I wonder about with mineral wool are related to the fact that its stiffness is more than an order of magnitude lower than the stiffness of the foams (Tables 9 and 24). That jives with the comments in this thread about it being difficult to get the strapping flat, and underscores the need for discussion of strategies such as the double-thread Topix screws and the original idea here--a plastic tube behind the strapping around each screw--to get it flat. However, the conclusions about the mechanisms of vertical displacement resistance point largely to mechanisms that actually rely on compressing the insulation to the point where it develops significant pressure against the wall and the strapping. At least one photo makes it look like that is the way your testing was done. So I worry that if people use strategies that avoid compressing the mineral wool, they may end up without the benefit of those vertical displacement avoidance mechanisms. Friction seems like a particularly dicey one to rely on, given that the particular materials involved might vary according to what WRB people use.

    I am left thinking that to really make a thick mineral wool system that is easy to install and certain to support siding well, I would want either:
    a) A system that makes it easy to anchor the strapping at a consistent distance out from the sheathing, without compressing the insulation, AND a way to provide additional vertical support beyond the cantilever bending of the screws, such as some triangulation in some of the screw attachments, or support at the top or bottom of the strapping.
    b) A system that makes it easy to anchor the strapping at a consistent distance out from the sheathing, with the insulation compressed a consistent amount.

  25. John Straube | | #25

    Response to Charlie Sullivan
    Hi Charlie,
    I think you don’t have a clear understanding about how science informs engineering, and then how engineering turns into building design. You seem to think we need to know everything about a material or system to use it. We don’t, and we can't.

    We have not yet got a good predictive method for predicting the pullout of nails from studs. When you test, you get a larger scatter, larger than we get when we test insulation holding up cladding! So papers and reports are being written that talk about all the unknowns and the research we need to do to improve our understanding of nail withdrawal. That does not mean we can't build with confidence based on the information we do have and the experience we do have. We have been building millions of homes without precise and accurate information. All we needed was sufficiently good/ reliable/safe info.

    While researchers will be trying to assess how much resistance is provided by strut and tie (the majority, but is it 80%, 65% or?), friction (should we assume zero? I do) and screw bending (10%? Not worth considering, ignore it), you can use conservative design values and it won't affect the design. Eg. #12 screws, 16” o.c. will support all lightweight claddings through reasonable thicknesses of insulation (perhaps 12” is no longer “reasonable”). None of the uncertainties have any practical meaning at that point (unless you somehow think you can stretch the spacing to 24x24” and want to use #8 screws through 12”). There is not need to have any meaningful amount of compression, just be in “substantial contact” with the insulation. Hard to know how you would be able to even not build it that way.

    You say insulation has an order of magnitude less compression resistance. True when compared to a standard test method, but it does not appear that way when installing it. I personally warn people not to blindly chose the highest stiffness Roxul insulation because I have used it several times and find it TOO stiff and unyielding. I like more give so that I can get the insulation in contact with the substrate, and so that I don’t have to cut the insulation as tightly to get butt joints. But you may like that hard insulation. After five projects, you can report back.

    While getting the furring strips straight is definitely something to think about, this can be a problem with EPS insulation as well. It has just not been a that big a problem in practice. Remember, hundreds, perhaps thousands of houses have been built this way in the last 5 years. Perhaps people experiencing problems are not using screw guns with clutches (a couple of YouTube videos of the Heco screw show that!) or large pancake head washers (which don’t allow the compression of wood accommodate the last 1/16” of screw without major compression of insulation).

    I don’t doubt that there are people who want what you want: a system that ensures the space between sheathing and furring is tightly controlled to the nearest 1/16”, or perhaps a system to avoid compressing the insulation. But these are not technical requirements to build a wall, they are personal desires, and if there is a cheaper, easier way to build, the market tends to focus on that. Your desires can be met by a growing range of systems: they all demand a premium of $1-3/sf over the screw-only solution and they all perform worse thermally and take more effort to install. Check out the commercial systems at:

    The primary reason to use these alternates is either: 1. A very heavy cladding, 2. Very uneven backup requiring adjustment of more than ¼"-1/2", 3. Or attachment in specific locations hard to target when insulation is attached.

    There are many practical situations when clips like Knight Wall, Cascadia, Smart ci, are the best solution. Usually these are in commercial situations with taller buildings, worse tolerance, expensive and heavy claddings. For your typical single-family home with common siding, I can't see the need.

  26. Expert Member
    MALCOLM TAYLOR | | #26

    By any chance have you worked out what using #12 screws @ 16" o.c. would cost on your place?
    Given that we are also discussing an alternative double-wall in another thread, I wonder how the two compare in our region? Even accounting for the extra labour of framing the two walls separately as suggested by Kevin's detailing, I bet it is substantially cheaper than the exterior mineral wool.

  27. charlie_sullivan | | #27

    I can agree to a simple solution!

    Thanks for the prompt and extensive reply to my concerns. I think my description of wishing for a way to achieve consistent compression may have implied a need for more precision than I intended. It sounds like a screw gun with a clutch is good enough to achieve sufficient and sufficiently consistent compression. That's a lot easier than cutting lengths of PVC pipes and holes for them in the insulation, and a lot cheaper than a commercial attachment system.

    That sounds good to me--it will be interesting to see if reports from the field confirm that it's easy to make things adequately flat that way. I get the impression from the discussion here that some people find it difficult, but as you note that may be people who use screw guns without clutches set appropriately. On the other hand, if, as you note, it can be difficult with EPS, it certainly seems like there's room for it to be more difficult with mineral wool, and therefore room for people to come up with better systems to make it easier with mineral wool.

    I also didn't mean to imply that we'd better have it all figured out and understood scientifically before we can build anything. My primary hope is to avoid the waste of time and money that occurs when things like this get figured out again and again for each project, by having clear assurance that a simple system works. That assurance can come from a combination of experience and analysis. Recommendations to build a test wall make me think we aren't quite there yet. When I say we aren't quite there, I only mean that we have extra costs and doubts impeding projects, not that we shouldn't build anything yet.

  28. StollerB | | #28

    Cost Implications
    Malcolm- I have not yet worked out a cost comparison now that I have a better concept of what thick exterior mineral attachment demands. I have also worked out some solutions to my double stud build ability issues on our other thread. Once I do a quick cost analysis and get some pricing on other Roxul products that John has suggested as alternatives, I will report back on comparisons.

  29. charlie_sullivan | | #29

    Another report ... this one with specific recommendations
    I just read another excellent BSC/BA report on this:
    Initial and Long Term Movement of Cladding Installed Over Exterior Rigid Insulation
    It's labeled as a draft, but it's 2 years old.

    Towards my hope of making this easy without a lot of churn before you can start building, it has a specific recommendation: a maximum of of 10 lbs/fastener for 4" thick insulation, based on #10 screws. That's converted to a handy table on p. 36 given the resulting fastener spacing based on cladding weight, typically 18" for 16" on center furring or 12" for 24" on center, with light weight cladding. Unfortunately, it doesn't give guidance for using thicker insulation.

    Interestingly, it seems to have a different take on the mechanism than John does in comment #25. John recommends not bothering to count any mechanism other than strut and tie, but the report's discussion of the strut and tie effect shows that its vertical load bearing contribution is theoretically quite small and hypothesizes that it's only important in adding to the normal force to enable greater friction.

  30. GBA Editor
    Martin Holladay | | #30

    OK - Who's going to build the mock-up?
    If a GBA reader ends up building a mock-up with 6 inches of mineral wool and furring strips:

    (1) Post a comment here telling us about your plans.

    (2) Take photos.

    (3) Report your conclusions here.

    That way, it won't be necessary for 10 GBA readers to build 10 mock-ups.

  31. Irishjake | | #31

    Screw It!
    For Tru-fast 9" SIP-TP screws I'm paying $100/250 screws, $126/250 (12"), and $223/250 (18").

    When you break down the costs of nails, wood,labor, etc., I'm not sure double walls are cheaper.

    Sourcing the types of Roxul John described proved difficult for me last year (Spring 2015), maybe it's easier to source those materials now.

    Tru-Fast makes screws up to 18", that works great for exterior insulation (nearly identical to what is used in the CCHRC/ Remote Walls videos.

    Spax also makes a screw up to 23". They make screws very similar to Tru-Fast up to 18", and then from 18"-23" the screws are much larger in diameter.

    Heco makes screws to 18" too. Much stronger than most all domestic screws. They are used routinely in Germany for exterior insulation installations that consist of woodfiber insulation with thicknesses that are much larger than 12".

  32. Irishjake | | #32

    Even if it works, does it?
    If a tree falls in a forest, and there is no one around to hear it, does it make a sound? - This whole bit on attaching exterior insulation to walls and roofs is a bit metaphysical. We keep asking, "Does it work?", but why do we keep asking if as John mentioned it has already been done. The answer is already there. In practicum, installing thick layers of exterior insulation has been successful, both with Roxul and with rigid foam. The issue at hand though is not that there aren't projects with proven performance. The problem is widespread acceptance, and knowledge surrounding the strategy.

    Code enforcement officers and engineers only have the data provided by varied screw manufacturers, and a few papers by some forward thinking building scientists to go by. Telling a code enforcement officer or engineer, "It worked for them, it will work here.", does not get your permit approved. It does not satisfy the human desire to know why. It does not satisfy insurance companies, lawyers or judges.

    I have spent hundreds and hundreds of hours contemplating my exterior insulation strategy. I went around and around in circles. Certain engineers told me it wouldn't work, or shouldn't. Now what the heck do I do? I mean the best of the best building scientists, engineers and architects say it will work, and have shown that it will. If this is the case - Why isn't thick exterior insulation prescriptively outlined for all to use? Why is it so hard to find building details for thick exterior insulation on walls, above doors, windows, garage doors, on roofs, at the ridge, or where a roof meets a wall?

    So even when I know it works and I've seen pictures of it working - it doesn't answer my over excessive tendency or hunger for the knowledge of why.

    It is reverse psychology - in the real world, I'll take someone who knows and can apply the knowledge any freaking day over someone who has a piece of paper that says they know, but don't actually know jack. In this situation - we've got the real world examples to show us it works, but here we are searching for the little piece of paper to say - "hey, this works!".......

    What. The. Heck!

  33. StollerB | | #33

    My Takeaway Summary
    After all the reading of reports, etc, here is my take-away data from all of this to guide me in detailing future projects. I know that some of my guidelines noted at the end may deviate from what differing individual reports may say, but overall, this is what I think to be a safe practice for the different cladding weights. Your takeaway/comfort level may be different :)

    Exterior Roxul Attachment

    Study References:
    - BSC 3” Comfortboard Attachment Report:
    - BSC 8” Roxul Rockboard 80 Deflection Report:


    Roxul Rockboard 80 Data Sheet:
    - Density: 8lbs/ft3
    - Compressive Strength: 439 psf @ 10% compression
    - Thickness: 1” - 5”
    - Dimensions: 24x48, 48x72

    Roxul COMFORTBOARD IS (80) Data Sheet:
    - Density: 8lbs/ft3
    - Compressive Strength: 439 psf @ 10% compression
    - Thickness: 1.25”, 1.5”, 2”, 3”
    - Dimensions: 24x48, 36x48, 48x72, 48x96

    Roxul CavityRock DD (dual density) Data Sheet:
    - Density: Outer Layer=6lbs/ft3
    Inner Layer= 4.1lbs/ft3
    - Compressive Strength: 439 psf @ 10% compression
    - Thickness: 1” to 6”
    - Dimensions: 24x48, 16x48

    GENERIC APPROXIMATE Cladding Weights (assumes inclusion of rain screen & fasteners): (psf)
    Vinyl Siding: 0.6-1.0
    Wood Siding: 1.0-1.5
    Fiber Cement: 3-5
    ¾” Cement Stucco: 10-12
    Adhered Stone Veneer: 17-25

    SPECIFIC Cladding Component Weights: (psf)
    Hardi-Plank: 2.5
    Hardi-Panel: 2.3
    1x3 wood furring @ 16" o/c: 0.35
    1x3 wood furring @ 24" o/c: 0.24
    1x4 wood furring @ 16" o/c: 0.48
    1x4 wood furring @ 24" o/c: 0.32

    (The below guidelines should all ensure vertical deflections of less than 1/8” over long term durations.)

    For TOTAL insulation, rain screen, fastener and cladding loads UP TO 15 PSF:
    - Any of the above Roxul products
    - 1x3 strapping at max. 24” o/c into framing
    - #10 screws w/ min. embeddment of 1.25” into framing
    - fasteners spaced max. 18” o/c vertically

    For TOTAL insulation, rain screen, fastener and cladding loads UP TO 30 PSF (and to attempt to prevent the likelihood of cracking in stucco, which cannot be guaranteed!):
    - Only COMFORTBOARD IS(80) or Rockboard 80
    - 1x4 strapping at max. 16” o/c into framing
    - #12 screws w/ min. embeddment of 1.25” into framing
    - fasteners spaced max. 12” o/c vertically

  34. Expert Member
    MALCOLM TAYLOR | | #34

    I was specifically wondering about the combined cost of long fasteners and exterior mineral wool insulation. One advantage of double walls is they generally don't have expensive components. If it's a wash, then it largely comes down to preference. But a substantial cost difference one way or another may make a compelling case for either approach.

  35. Irishjake | | #35

    Burke - strong work

    Strong work - great summary!

    One question - where did you find the spec or guideline to only embed the screw 1.25" into the framing? Seems in almost all my findings it has been a minimum of 1.5"....

  36. StollerB | | #36

    One of the BSC studies tested assemblies with different fastener embedments, some with each fastener through the sheathing and into the studs, some with all the fasteners missing the studs and only into the ½" sheathing, and some only into the sheathing except the top and bottom ones into the plates. There was really no difference in sag under hydraulic ram load testing of each of those scenarios, so from my perspective, 1.25" sounded like a fair embedment as a minimum. I agree that 1.5" would be even better, but it seems to be unnecessary according to the testing data in said report (can't remember which one of all the ones I read!)

  37. Irishjake | | #37

    I was using 1.5" as a guideline because it seems most screw manufacturers spec that depth. My engineer also spec'd that. I've seen it in other publications and studies too.. I passed over those embedment tests, so thanks for highlighting them for me.

  38. ryem | | #38

    Malcolm Taylor
    We use 2" 8'x4' sheets of eps and rip on a table saw. It's a dirty job, but not too difficult. We pre drill the wood studs, then insert screws through wood and foam being sure to keep them in plane edge to edge. I hold a speed square as I install the double studs to be sure they are set to the right depth of 7 1/4".

  39. charlie_sullivan | | #39

    Some clues on why the measured performance exceeds modeled
    Recognizing that we don't need to completely understand how something works to be confident in building it (see #25), and that we have specific guidelines on how to build, (see #30 and #34), it's still interesting and potentially useful to try to figure out some of the structural engineering puzzles left hanging in the reports and in the discussions above. These puzzles include:
    1) Why does the measured vertical load capability exceed the modeled capability?
    2) Why does the mineral wool seem stiffer in practice than the very low modulus measured would predict?

    In addition to the fun of figuring out a puzzle, some reasons to pursue these questions include avoiding unnecessarily overbuilding, and, perhaps more importantly, knowing how it works makes it easier to be sure a change that isn't expected to be consequential won't undermine the structural integrity.

    The attached plot is from Andrius Buska & Romualdas Mačiulaitis (2007), "The compressive strength properties of mineral wool slabs: Influence of structure anisotropy and methodical factors", Journal of Civil Engineering and Management, 13:2, 97-106. It shows the stress/strain curve for compression in three directions. Curve number 1 is the direction we are interested in, compression in the direction of the thickness of the slab. The samples were about the same density as Comfortboard IS.

    There's a knee around 5% compression, but of more interest is the fact that there's also some curvature at the bottom left, making the overall curve S-shaped. Presumably the 20-28 psi modulus number in the BSC report is just the initial slope. Beyond about 1-2% compression, the curve gets steeper, indicating greater stiffness. 2% compression of six inches of insulation is about 1/8". So in the range of up to 1/8" to 1/4" compression, we are in the range where it's quite stiff--the slope there corresponds to about a 180 psi modulus, 6-9 times higher than the values in the BSC report.

    But even that higher number doesn't result in significant vertical stiffness using the strut-and-tie model in the BA-1314 report. (The equations on p. 18, in which it's worth noting that for the Fc equation to work, delta should be a fraction of the thickness, not in inches, which isn't clear from the definition or the figure.) But that idealized equation misses a few things that can happen in the real world:

    1) There's some pre-load compression of the insulation. According the BSC's measurements, typically about 150 lbs per fastener. Adding that helps a lot--it changes the theoretical results from a negligible contribution from strut and tie to supporting 5 lbs/sf at 1/4" displacement, assuming 180 psi modulus in the working range, and 16" x 16" fastener spacing. We could also call it 2.5 lbs/sq ft at 1/8" displacement, and count on the other mechanisms to provide the other 2.5 lbs. We could call that good enough, but it's just barely good enough. A factor of safety might be nice. And the test results indicate that it's better than this prediction in practice. So there must be something else going on.

    2) The low contribution from the theoretical model results from the behavior of tangent and cosine functions near angles of zero. This means the angle must increase--and the cladding must shift down--before the strut force goes up and before the tie force includes a component upwards helping to hold the cladding up. But getting the angle of the screw right at zero would be incredibly hard to achieve in practice. If we assume the screw starts out angled just 6 degrees off, angled so it is up a bit at the sheathing, down towards the strapping, we can already support 16 lbs/sq ft at 1/8" vertical displacement, with 130 lbs of pre-load compression per fastener, at 16" spacing. That's all you need, even without counting any contribution from friction or the screws bending strength. So I'm guessing that the scatter in screw angles is what makes the test walls better than the predictions.

    So what does that mean in practice? Probably in most cases, the screw angle scatter and pre-compression will be all that is needed to get sufficient strength. But it's worth considering how one might fail to achieve this. The Heco Topix screws are designed to prevent the pre-compression. It's conceivable, though unlikely that either a uncannily skilled carpenter or a fancy future auto-leveling automatic screw-gun might reduce the scatter. Or, worse, someone might accidentally aim all the screws at a slight downward angle, which would of course reverse the effect and undermine not only the strut and tie mechanism, but also the friction.

    So it probably make sense to specify at least one of the following:
    A) Specify that the angle of the screw be zero to slightly upwards. If a particular installer can control the angle to within plus or minus one smidge, aim for one smidge upward so that the result is between zero and two smidges upward, and never downward at all.
    B) Specify that every other screw be angled upward something like 15 or 20 degrees.
    C) Add just one screw per strap angled up 45 degrees.

    B) and C) are probably overkill, but they are not expensive to do, so why not?

    I had previously suggested that C) would require co-location with something like a Heco Topix screw that would provide the strut function, on the assumption that the mineral wool would be too squishy to provide that function. But given the stiffening behavior that the curve here shows and that John reports experiencing in practice (comment #25), that's not really necessary.

    In summary, considering scatter in screw angles, pre-compression, and the nonlinear compression behavior of mineral wool can probably explain why the tests show better performance in practice than the previous models predicted. We can choose to use that information to assume it will continue to work fine, or we can deliberately angle some of the screws a bit in the right direction and have structures that will more easily support larger loads with less deflection.

  40. StollerB | | #40

    Some of that is a bit over my carpenter head, but your conclusions about screw angle scatter improving the results seem to make intuitive sense to me. And as you said, it's very easy to do on site, so why not? Thanks for the continued thought and response Charlie!

  41. user-4524083 | | #41

    Charlie - Thank you for that ! Does curve #1 also pertain to under concrete slabs? I'm considering using Roxul for this purpose.

  42. charlie_sullivan | | #42


    Yes, curve number 1 would be for stress in the direction of weight of a slab on sub-slab insulation. That curve is not specifically for the Roxul product--it just shows what a similar density material does. But it does give you a better sense of how it works--the weight of the slab would compress it the first 1 or 2%, but beyond that it gets stiffer so there's little reason to worry about it not having adequate support. Martin's blog on sub-slab mineral wool is a great reference for anyone who hasn't read that:

    The 745 psf spec on CIS at 10% compression corresponds to 0.0357 MPa on the scale on the plot I posted. We see that curve one is a little higher than that, whereas curve 3 is very close to that. It might that that this material is a little stiffer than Roxul CIS, or it might be that Roxul builds a little margin into their spec and a typical sample comes out closer to this curve than to their spec.

  43. Irishjake | | #43

    Thanks for all that - although you lost me a few times - it was very eloquently put and helpful too!

  44. Expert Member
    MALCOLM TAYLOR | | #44

    Great discussion
    In particular thanks to Burke and Charlie for spending the time and coming up with interesting and valuable insights. I'm very grateful to both of you.

  45. user-5714885 | | #45

    grade 8 bolts
    Here is a crazy idea for those with metal construction. Drill a hole * completely through* your furring, Roxul board, and the flange of the metal stud. Insert very long grade 8 bolt. Torque down the nut on the other side, seated inside of the stud.

    Yay, nay?

  46. StollerB | | #46

    Cost Comparisons
    As promised (post #29), I have done some cost comparisons on my future house plans to determine which wall system would be more cost effective for me: a super insulated wall using the exterior Roxul or a double stud wall system with dense-pack cellulose. The short story is that the exterior Roxul is, in my estimation, more expensive in both materials and labour.

    For my plans with approximately 2970 square feet of exterior surface area (I did not subtract window and door areas in either estimate, for simplicity's sake), here is the breakdown:

    Assume a cost baseline assembly of a standard 2x6 wood framed with exterior sheathing which is the main air barrier, no insulation and treated plywood rain screen strips (which are code mandatory in our jurisdiction).

    Here, then, are the MATERIAL ONLY cost premiums (in CDN$, at contractor prices, not including taxes) to upgrade to either:
    - a 2x6 wall @ 24" o/c with 6" of exterior Roxul Comfortboard IS (three layers @ 2") and 1x3 rain screen, OR
    - a 12" wide double-stud 2x4 wall @ 16" o/c with exterior gypsum drywall and a layer of ½" ply on the interior for the load bearing wall. (See attached PDF for a section of the double stud wall, since it is a more complicated assembly to describe).

    Cedar 1x3 rainscreen @ 24" o/c: $487
    8" long GRK fasteners through rainscreen @ 19.25" o/c vertically: $1685
    Roxu/Rainscreen/Cladding attachment engineering costs (required by Roxul for >4" insulation): +/- $750
    Roxul Comfortboard IS (three layers of 2"): $18192
    Cost of extended through wall flashings (6" deeper than normal): +/- $500
    Fibreglass batts for the 2x6 stud cavities, with a 6 mil poly vapour barrier, including installation: +/- $3680
    TOTAL= $25294

    Extra framing: $1046
    ½" Denseglass exterior gypsum: $2379
    12" Dense-pack cellulose (including installation): $13663
    Interior secondary air barrier air sealing with 3M All Weather Flashing Tape: $278
    Inner Wall "temporary" bracing during rough-in phase: +/- $250
    TOTAL: $17616

    MATERIAL ONLY COST DIFFERENCE: Double stud wall is over $7600 less expensive.

    That's not as large of a material cost difference as I had expected, but it is still worth considering. (It would pay for the purchase and installation of a high quality HRV system!) However, when I started considering the labour costs between the two systems, the exterior Roxul system really started to look onerous. To frame the double stud wall I am proposing would realistically add about 4 days max. to the wall framing of the house. That's pretty much the extent of the extra labour required for the double stud wall. Alternatively, the exterior Roxul attachment in 3 layers would take far longer, once one factors in the additional work surrounding how the Roxul terminates around window/door openings, how one installs 3 layers of Comfortboard concurrently without having them fall off, as well as having to drive 1150 (on my project) 8" long screws one at a time through the rain screen (and then checking it all to make sure it is flat/co-planer) instead of just nailing it to the framing with an air nailer, as well as a bit more time bending and dealing with the much longer through wall flashings (which would all be 6" deeper than those on the double-stud wall), and how those flashings interrupt your installation of the Roxul as you move above openings. As a ballpark, I would guess that labour for the exterior Roxul would add at least 12 days of extra labour at best. So, that's 4 days of labour for two guys vs. 12 days extra labour for 2 guys.

    Combined with the material cost differences, for my scenario I figure the double-stud wall all-told will probably be around $13000 less expensive AND a good week or more faster to build in terms of timelines.

    I understand that every project is different. Different climates demand different assemblies, and different material availability and trade expertise are all important considerations. But for me it was a valuable exercise to price things out and consider labour times to determine which wall system seems to make more sense from a purely financial perspective. While a fantastic wall system in terms of hygrothermal performance and durability, the exterior Roxul SUPER-INSULATED wall system is definitely an expensive one.

  47. Rick_Milne | | #47

    Roxul Comfortboard 80 engineering report
    Could John Straube or Martin refresh or check the link to Roxul attachment report in comment 20 as it is not working and I want to ensure my engineer in BC reviews their latest and best advice!!

  48. GBA Editor
    Martin Holladay | | #48

    Response to Rick Milne
    Roxul seems to have removed that report from the company's web site. Can any GBA readers help Rick Milne?

  49. GBA Editor
  50. Rick_Milne | | #50

    Response to Martin
    John refers to a table 4 and deflection results in his comment 20 I think it is a separate document. I called Roxul about a month ago and they referred me to their fastener guide which indicates that a horizontal single screw through strapping is permitted, but the guide is a bit nebulous on engineering reports to support this. I am trying to give my engineer some info as he has me going down the two screws with one at an angle road. I also have a Cascadia Clip I am considering but am hopeful that something from the manufacturer may sway my engineer and building inspector.
    Many thanks for trying....I will also email John in Ontario.

  51. Rick_Milne | | #51

    Roxul Comfortboard Attachment Update
    Martin, I heard from John and wanted to share with any other GBA readers...Here is the latest info from Roxul:

    As John also mentioned in #20 above there is an article from Building Science that supports the use of regular construction screws as opposed to specialty screws outlined in this latest report from Roxul.

  52. GBA Editor
    Martin Holladay | | #52

    Response to Rick Milne
    The link you provided doesn't work. But I did my own searching, and I think I came up with the right document and the right link. How's this?
    Cladding Attachment and Support

  53. Rick_Milne | | #53

    Response to Martin
    That’s it !!

  54. chimewind | | #54

    Thick Insulation Test
    To cross-post from another thread with information that is relevant to this one.. See this study released late last year:

    Basically showing that even at 12" of Comfortboard 80 there was not an issue with deflection using 8mm screws.

  55. GBA Editor
    Martin Holladay | | #55

    Response to David Bainbridge
    I have added links to resources from Roxul, Building Science Corporation, and Building Science Labs to the "Related Articles" boxes of three or four GBA articles on exterior mineral wool insulation. Thanks for the nudge.

  56. Romulus_5000 | | #56

    Zee purlins. Or channel purlins. Simple.

Log in or become a member to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |