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

Maximum Insulation in Thinnest Wall

J James | Posted in GBA Pro Help on


I live in downtown Toronto (Zone 6), Canada. I’m building a brand new addition to an existing 1888 brick house, on a narrow 17′ wide lot. I’m interested in getting the best thermal performance (R40) within the thinnest assembly (11″ wide) so here is my idea:

From inside to outside –
5/8″ GWB
2×4 wood stud framing with
3.25″ of high density spray foam (R21)
1/2″ plywood
air barrier
5″ Roxul Mineral Wool (R20)
1″ wood battens or alternately 6″ thermally broken z-girts
metal cladding (I am hoping to use a flat seam metal panel of some sort.)

Any thoughts would be much appreciated. I want to make certain that I haven’t missed any major holes with the thinking. I’ve been reading this forum for a little while now and am impressed by all the curiosity, knowledge and wisdom.

Also, do you think I need a vapour barrier, considering the spray foam acts like one? If so, should I locate it under the drywall?


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. GBA Editor
    Martin Holladay | | #1

    You are aiming for R-40 in 11 inches. That won't be too hard, as long as you pay attention to thermal bridging. All you need is R-3.6 per inch, which is fairly easy to achieve.

    There are a great many ways to achieve your goal, and your present plan will work.

    Q. "Do you think I need a vapor barrier, considering the spray foam acts like one?"

    A. As you guessed, closed-cell spray foam is an excellent vapor retarder, so you don't need an additional membrane. For more information on this topic, see Do I Need a Vapor Retarder?

  2. J James | | #2

    Thanks Martin,

    Yes, R-40 in 11 inches, I forgot to mention that I was also trying to keep the costs down. : )

    Great advice and blog, by the way.

  3. Expert Member
    Dana Dorsett | | #3

    Pitch the closed cell foam in the cavities- with the thermal bridging it's buying you less than R1 in whole-wall performance over cellulose or open cell foam. Only a spray-foam viruoso can really get a guaranteed 3.25" of closed cell in a 2x4 cavity because it's damned-near impossible to trim without messing up the studs (unlike open cell, which trims easily.) But assuming you really can at R6.5/inch, at a 20% framing fraction that R21 center-cavity is only ~R10 (not counting sheathing & gypsum) after thermal bridging, which is less than half the center-cavity value for that layer(!).

    A full cavity fill of R3.6/inch cellulose or open cell foam comes in at ~R9.3, due to ~R3.5 compared to ~R3.25 for the framing fraction.

    How much are you willing to spend for R1? For cavity fill go with either cellulose (which is protective of the wood by it's moisture buffering characteristics), open cell foam (which makes air sealing the assembly much easier), or dense packed(1.8lbs per cubic foot min) blown fiberglass (L77, Optima, Spider, etc.), which is somewhat higher-R. Damp sprayed cellulose is usually (but not always) the cheapest in my area, dense-packed fiberglass is usually more expensive than open cell foam, but local markets will vary a lot on this.

    From a dew point control at the plywood air-barrier sheathing point of view, R20 of exterior Roxul is plenty for your stackup in a Toronto climate- you don't really need a vapor retarder stronger than standard latex paint on a nominally air-tight interior gypsum layer. If the inspectors balk at the no vapor retarder concept, you can use "vapor barrier latex" painted on the interior side of the plywood (or on the gypsum, if the inspectors really REALLY insist.)

    To hit closer to true R40 whole-wall you can replace the 5" of Roxul with rigid polyisocyanurate foam. Virgin stock is expensive, but if you used at least 4" reclaimed roofing foam from commercial re-roofing & demolition, with maybe the outer 1" foil-faced virgin stock for ease of air-sealing it's usually cheaper per unit-R than Roxul. (Call some commercial roofing contractors, they will often have a stash of scavenged goods. will ship nationwide in the US, not sure if they ship to Canada, but shipping adds to cost too- local is better/cheaper.)

  4. J James | | #4

    Thanks for the tips, Dana. I was concerned about the thermal bridging as well. Just some clarifications, please:

    1 - Wouldn't the thermal bridging from the wood framing reduce any kind of insulation value by the same factor?

    2 - Why would the polyiso be considered a closer "true R40" vs the mineral wool?

    I'm going to look into the scavenged goods idea.

  5. Expert Member
    Dana Dorsett | | #5

    In order:

    1: The R-value of the framing timber runs about R1 for every inch of thickness, so you're at about R3 for 20% of the surface area of the wall. (That's a fairly conservative framing fraction too, which takes some careful design to hit. Framing fractions of 25% are more common with 16" o.c. stud spacing and doubled-up top plates & window headers. Where seismic reinforcement and fireblocking are required by code it hits or exceeds 30%.) That R3.5 doesn't change with the R-value of the 80% of the area that is cavity fill, but the DIFFERENCE between the cavity fill & framing fraction goes up with high-R foam.

    And where the cavity isn't fully filled, in a first-order 2-D approximation, only the section of the stud with the cavity insulation counts- the protruding section of the stud is at room temp. So with the only 3.25" cavity fill you are losing about R0.25 on the framing fraction. (In a 3-D heat transfer model it's slightly worse than that, but it's "in the noise" on the R-value of real studs' performance at different moisture contents, etc, a "don't care" level of difference.)

    The average R- performance of the wall is a function of it's average heat transfer rate or U-factor. By definition U-factor == 1/R (and conversely). With 20% of the wall at R3.25 and 80% at R21 the average U-factor is 0.2/3.25 x 0.8/21 = 0.0996. Converting that back to an average R value (R=1/U) R= 1/0.0996= 10.037

    If it's say, R3.6/inch open cell and a full fill you get a center cavity R of only R12.6 (how HORRIBLE! :-) ) but the 20% stud fraction rises to R3.5. The average U-factor of the assembly at a 20% framing fraction becomes:

    U= 0.2//3.5 + 0.8/12.6 = 0.1206

    And the average R is then 1/0.1206= 8.3
    (^^ oops- note the typo in the stated R in my prior post where I called it R9.3. Mea culpa! ^^)

    So in the closed cell case, despite a center cavity R that is about R9 higher than with an open cell cavity fill, it only buys you an R1.7 improvement. (That's a fairly expensive R1.7 too!)

    The moral of the thermal bridging story is, save the high-cost high-R/inch goods for the exterior of the sheathing, where it's performance isn't being cut of at the waist by thermal bridging of the structural timbers!

    Which brings us to...

    2> Poyliso is rated at 75F at about R6/inch to rock-wool's nominal R4/inch. A 5" layer of polyiso is nominally R30, the studwall is about R8.3, which brings it to about R38.3, add another R1 for the gypsum and sheathing and you're at about R39.3, which is pretty close to R40 whole-wall, eh?

    But in fact the mid-winter performance of the exterior iso is lower at lower temperature- a characteristic of the foam- some foams increase in performance at lower temp. In Toronto's climate would really only average something like R5.7/inch, bringing the foam layer down to "only" 28.5, so it's really more like ~R38 than ~R40 during the time where it counts the most, with a heat loss about 5% higher than a true R40 wall. (That's still a fairly high-performance wall though.) In Winnepeg's climate the mid winter performance of that stackup would be lower still, at a bit over R37. (Aren't you happy to be living an a "warm" climate? :-) )

    Rock wool's performance increases slightly at lower temps, but not enough to even hit R4.5/inch at the temperature extremes, let alone iso's derated R5.7/inch.

    When using an exterior continuous insulation approach, it's important to keep the thermal bridging of the fasteners as low as possible. The furring/girts need to be screwed to every stud with a timber-screw, but if the siding- flatness can tolerate it, keep the spacing between girts at 24" or higher. Steel screws have on the order of 1000x the thermal conductivity of foam or rock wool insulation, so it doesn't take a huge cross sectional area to make a measurable dent in thermal performance.

  6. J James | | #6

    Got it and got it. Thanks Dana. This has helped a lot. (Coincidentally, I just saw that little caveat with the rock wool performance relative to outside temperature in the product literature. : ) ) Time to rethink...

  7. Expert Member
    Dana Dorsett | | #7

    A 6" steel stud? Surely you're joking!!

    Steel studs are a HUGE thermal bridge with conductivity orders of magnitude higher than wood! That would be a net DROP, not a gain in performance.

    I was assuming the Roxul was continuous insulation, with the support for the 1x or girts being widely spaced screws/bolts or something, but the details of how you do that IS a serious consideration for the thermal performance of the assembly.

  8. Jerry Liebler | | #8

    NO!!!!! PLEASE read what I wrote! ROXUL FOR STEEL STUDS 16" OC placed between custom ripped 6" girts of WOOD on 17 1/2" centers! Yes the wood is a thermal bridge where vertical studs intersect the horizontal girts but I discussed that. too

  9. Jerry Liebler | | #9

    The higher density Roxul is closer to r 4.2/" at 70 f so it just might hit the r 4.5 and softwood framing lumber, according to the lumber industry is 1.25/", using Dana's value of 1/" forces you to minimize wood which is good and his value gives a very worst case estimate. Also using high density Roxul can be almost as costly as foam, The rigid boards sell for more about 3 times the $/r sq ft of .comfort bat. Roxul comfort bat for steel studs is available in a 6" r24 product. I assume your steel cladding requires horizontal members to mount. Since this is a small project it may be worth while to rip 2x8 to a 6" width and attach them horizontally, with screws through the plywood from the inside on 17 1/2" centers so you can fit r24 bats for steel studs between them creating a,sort of reversed Mooney wall yes you would have 1 1/2" square thermal bridges at the intersections but the total heat loss would be very small as the bridges occupy about 0 8 % of the area. Let's assume r 14 Roxul in the cavities, the battens are 1 1/2" wide and 17 1/2" centers and only horizontal so they occupy 8.6% of the area and 25& framing for the 2x4 wall meaning 25% of it's area is wood.
    We really have a 3 dimensional heat flow problem and in addition to the heat flow through the wall heat will be flowing up,down and sideways in the plywood which has a thermal resistance of 1.25/" in all directions. We can bracket the actual situation with 2 easy cases. one where the plywood is a perfect insulator and the other where it is a perfect conductor The perfect conductor gives a limit of r29, the perfect insulator gives r41 the actual 3 dimensional heat flow lies between these limits. Experience suggest it is closer to the upper but without an elaborate simulation we only know the limits. In any event that's about the lowest cost. Using Roxul's AFB instead of comfort batt would gain a little but cost about double. if your siding will allow it you could use 6" (edit THE WORDs "ROXUL FOR" SHOULD BE HERE) steel studs for 24" centers and gain maybe r1 .
    EDIT After a little more thought with the "system" I described you'll have condensation on the inside of the plywood where the ripped 2x8 runs and the only answer is go back to the spray foam for the inside, even though it adds little r value and costs a good bit more.

  10. J James | | #10

    Thanks Jerry.
    Further to your thoughts, I was wondering about doing 2 layers of 3.5" Roxul Steel Stud rock wool, switching the framing direction from vertical to horizontal to minimize the frame thermal bridging but then I think I'm starting to get into strange framing techniques and thicker walls and at that point, I might as well stick with the polyiso or the higher density rock mineral as the sheathing insulation...

  11. Richard Patterman | | #11

    Maximum insulation in thinnest walls: why not polyurethane SIP?

    They are expensive but when you need to maximize space on a small project they make sense.

  12. J James | | #12

    Hi Richard
    I'm building on the lot line on one side and then 3' from the other so both walls need to have 45 min fire rating and I don't think SIP will pass. However, I was thinking about them for my cathedral ceiling. Any suggestions about manufacturers and any idea of a sf budget price? This is a small addition, !2' x 12', do you think they come for such small projects?

  13. Richard Patterman | | #13

    I'm sure you can get a 45 minute wall with exterior gypsum board and hardcoat stucco.
    My source is in Colorado, you would need to find someone near your job.
    You can build ant size project with SIPs. I have a friend that builds large dog houses and campers.

  14. J James | | #14

    Thanks Richard. I'm at the beginning of planning so there's a long road ahead. : ) I like the thought of a SIP dog house!

  15. Expert Member
    Dana Dorsett | | #15


    My-bad! ( I should read more carefully, eh? ;-) )


    Why not polyurethane SIP?

    OK, here's the nutshell version...

    Most closed cell polyurethane foam is blown with HFC245fa (or worse), with more than 1000x CO2 global warming potential (GWP). In high-R assemblies the lifecycle environmental hit of just the blowing agents (never mind any polymer manufacturing related issues) far exceeds that of the energy use it is offsetting.

    Honeywell recently release a low-GWP blowing agent (trade name "Solstice"), and DuPont has a low GWP version (FEA-1100), that may or may not be released yet, but these are VERY recent developments, and market penetration will take time. There are also at least two manufacturer's with water blown closed cell foams, Aloha Energy's 1.8lb foam (~R6/inch) and Icynene's MD-R-200 (~R5/inch), neither of which are going into SIPs, which are usually higher density/ higher-R.

    Until a polyurethane SIP manufacturer is specifying that they use only <7x co2 gwp blowing agents on the foam, they will be more environmentally damaging r for than rigid polyisocyanurate (blown with pentane at 7x gwp.) i've yet to see any mention of that polyurethane sip manufacturers' literature, but i'll thrilled when actually happens. until then i'm steering well clear them.

    It may take a Montreal Protocol type of regulation for high lifecycle-GWP building materials to finally make that change within the industry. While HFC245fa & HFC134a were a HUGE improvement over the CFC/CFHC blowing agents from an ozone depletion point of view, they're still gonzo-bad stuff from a greenhouse potential perspective. Blown with HFC245fa (as most closed cell polyurethane in the US is) it (and XPS) are the antithesis of '"green" even at code-minimum R-values, and are best used only sparingly as insulating-moisture-control layers, not as the primary insulation in high-R assemblies.

  16. Jerry Liebler | | #16

    Take a look at Cascadia clips as a method of mounting your external Roxul and cladding.
    They are fiberglass stand offs that hold a metal z bar to implement the rain screen gap Or maybe they are what you meant by thermally broken z? Their cost is low enough that one would be $ ahead using them and Roxul comfort bat FOR steel studs instead of high density Roxul and struggling to get the furring strips even. Using them aligned with your studs would allow the use of Roxul in the 2x4 cavities instead of spray foam as the condensation problem is far less.

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |