Foundation wall exterior insulation

Bright yellow rocket engine under the house Creeping cold on the slab edge
Rocket engine Creeping cold  
Recall these two infrared pictures, from part 24 when beginning to explore thermal imaging and then part 27 near the end?  These were among the eye-openers to the fact that my foundation wall was still a problem as far as overall heat loss. 

[Most images are linked to larger copies.]
Wall assembly before Don't get me wrong here -- the wall was generally in good shape, having gotten all air-sealed at the sill area and reasonably insulated and moisture-proofed on the inside with the extra layer of polyiso foam and trimmings, and there weren't really any obvious leaky or cold areas.  But even given the nominal trickle of heating it got when the air-handler ran, the basement stayed about 10 degrees lower than the main floor above it all winter -- and a delta like that across maybe R-2 of wood floor and carpets was clearly taking its toll on keeping the overall heat in.  I had become pretty convinced that despite my upgrades, the basement was still the source of the nonlinearities I was seeing in my BTU/HDD calculations, and I still wanted to try and do something to remediate that.

Basement heat loss concepts
In studying the problem a little further, I found this very indicative pair of diagrams in a builders' guide about foundations from a research team in Kansas.  Note that they're mostly talking about completely uninsulated basements and slabs, so I was already doing a bit better with all the added stuff on the inside.  But this basically illustrates the [lack of] R-value in concrete and soil.

Basement isothermal lines More insight came from a frost heave study and how foundation walls are affected, from our friends at Building Science.  This diagram [which I've taken a couple of liberties with] shows rough isothermal lines, i.e. the temperature gradient, between the outdoors and a basement insulated in a similar fashion to mine -- walls done on the interior, but not the slab.  The rocks in a rubble wall are thermally more conductive than soil and help pull more heat out of the ground and throw it away outdoors, with the effect that the cold can penetrate even deeper into the dirt near it.

My discussions with some of their folks a couple of years back had suggested that visible heat from a basement wall could easily be coming from a combination of the basement and the generic earth-temperature of the soil around it, while all that mass cools down over the course of a winter.  We do have to think fairly long-term about all of this.

Heat loss enhanced via cinderblock voids But my wall isn't rubble or solid concrete; it's cinderblocks.  Which have a stacked series of air voids inside, all the way up and down.  Some might believe that those air spaces would help insulate, but I seriously doubt that in this particular setup.  Here's my theory on why it might contribute to more earth-mass and footing heat loss: with a fairly generous area exposed above grade, the top part of the wall would easily chill down close to ambient temperature against the insulation on the inside, and likely take the sill down with it.  The difference between that and [relative] ground heat near the footing would set up convective loops through the column of voids, essentially helping conveyer-belt heat and cold exchange and contribute to chilling down the footing, the connected slab, and mass of earth around and underneath more than a solid wall would.  Basically, a flow of coolant with a big radiator, like in a car engine!  What I needed to do was slap a piece of cardboard over that radiator and try to keep the heat in, as much or as little as that might be. If nothing else it would help frost-protect the masonry, which was now running much colder in winter after re-insulating on the inside and probably imposing a larger delta across *all* of that modest R-15 top to bottom.

Not being able to measure any of this but only guess at a visualization from the evidence touched on above, maybe it was time for another building-science experiment.  On a fairly large scale.

  The exposed part of the wall was bad for heat leakage but great for moisture control, as it would let ground dampness out through the concrete face long before it could drift up as high as the sill.  With a solid vapor barrier formed from the configuration of materials on the interior, anything placed on the external surface would *have* to remain completely vapor-open or it would spell doom for the wood above.  That meant no foam, no membranes, and likely no paint.  The semi-obvious answer was rockwool, which the evident brand-leader Roxul makes in various types of rigid boards as well as the traditional soft batts.  I figured that covering the exposed parts with some meaningful R-value and going some nominal distance into the ground would at least help the wall stay a little more up toward the temp of the surrounding earth instead of the winter winds howling past it.

I started doing a little more research, mostly reading about Roxul's product offerings and application notes.  But the rigid board products aren't just something you'd find in the back aisles of the Big Box; they're somewhat specialty items and most are geared more toward the commercial market.  To know what might work for my not-yet-a-project I wanted to get some of the stuff in my hands, and thus began the first of my many runarounds with Roxul's customer service.  After a bit of legwork I had the number of a local rep in Rhode Island, who said he could send me a small batch of "hand samples" to look at and assured me that some of the products I was considering were ideally suited for what I was thinking about.

Spread of Roxul samples What I didn't expect was the two largish boxes that showed up a few days later, containing MANY nicely packaged samples of Comfortboard, Rockboard, BlackMat, and a couple of other things.  The Comfortboard, in its IS and CIS variants, seemed like the suitable candidates -- quite dense and rigid, and available in 3x4 foot sheets which seemed perfect for the areas I wanted to cover.  Two-inch thickness was manageable and would give me on the order of R-8 outside the cinderblock, which doesn't seem like a whole lot but would be far and away better than having it butt-naked to the weather.
  I also looked into adhesives.  While the Dynaflex caulk worked well on various kinds of foam it didn't seem appropriate for the more porous and somewhat friable surface of rockwool.  I needed some serious construction adhesive that would work on concrete and various kinds of stone, because basically I'd be sticking rocks onto other rocks.  I had some experience with the PL series from Loctite, and spending a while reading labels in the caulk and adhesives aisle seemed to suggest their "8X Fast Grab" product as the best option.

A little experimentation with that and some other sticky stuff revealed that the bonding strength of the rockwool board by itself wasn't actually that good, but I had already thought of a hack to improve that.  Basically, to prevent a small patch from ripping right off the surface, more of the fibers around a glued area need to be brought into structural play.  Now, what product did I already know about that could soak way into a porous substrate and then solidify into a strong supporting matrix?  Wood hardener.  I was pleased to find that it would create a much larger and more rigid region around wherever it was applied to the surface of the rockwool. 

In the foreground above, therefore, is my research tool: a small piece of cinderblock with little rockwool cubes glued to it, with and without the wood-hardener trick, that I set up and left to cure for a couple of days.

Test-patch layout The Roxul rep had sent along so many sample pieces, I decided that here was an opportunity to do a full test patch on the exterior wall by simply butting it all together and sticking it on.  I templated up a layout of mostly 2-inch stock and included a spare piece of XPS foam just to get a larger overall area, because I also wanted to see what covering just this much might do to the thermal profile of the wall and surroundings.

Bonding test block The next step for my glue-block was the "pull test".  As expected, the pieces without the hardener weren't that difficult to rip away, but the pieces *with* hardner took significant force that broke apart much more of the rockwool board in the process.  And these were just little dabs of the 8X adhesive, not the major long fat beads I'd expect to have running down the wall in a real installation.  That's what I wanted, so I had my mounting method at that point -- no messing around with the hammer drills and Tapcon screws everyone else was talking about.  I'd avoid weakening my wall and adding more thermal bridges and probably get a more airflow-resistant setup in the process.

Test patch going on So now it was time to take the whole concept outdoors and put it through some real-life trials -- bonding robustness, thermal effects, and long-term viability.  One contributor to the forums had mentioned another external-Roxul project going onto a home that he'd observed birds picking away at before the outer cladding went on, so one concern was what various critters might do to freely exposed rockwool over different seasons.  I already knew that insects avoid fiberglass and wouldn't tunnel through it, unlike any sort of foam in contact with the ground, but birds?  Squirrels or chipmunks?  Or would the rockwool just become hopelessly waterlogged and be no better an insulator than the concrete and dirt it was touching?  I mulled over various vapor-open possibilities for adding protective covering above grade, but maybe a sufficiently dense medium wouldn't really need it.

Well, I wouldn't know until I had bodged all this together and left it to the elements for a nontrivial length of time.  The major interest had to do with the fate of Roxul at and below ground level, so I placed the foam piece filling in along the top.

Viewport into test patch On a cold evening a while later, though, I got a pretty good hint about the effect this might have.  I left one of the sample blocks in the middle of the assembly unadhered and just held in by friction against its neighbors, with a couple of long screws sticking through from the back so I could gently slide it out and expose the wall behind.  When I did that and shot the infrared, I could clearly see that the test patch *was* keeping the wall a bit warmer.  Not fully up to earth-mass temp, certainly, especially with all the rest of the cold exposed wall not that far away through the abysmal R-value of concrete, but definitely helping protect it somewhat against the ambient.  And I *thought* I could see a slight difference up in the basement joist bays that were behind the test patch as opposed to the ones outboard of it.  Hard to tell there, however, as those joist bays have the disadvantage of the flashing-metal "heatsink" connected to the other side of the sheathing that forms their ends.

    Time passes ...

  I left the test patch in place and did nothing further on this for about a year, over a winter and summer cycle, and simply observed.  The timeframe on the entire idea was long and governed by cautious research; I was in no particular hurry for answers and had to be sure, knowing that to go beyond this and wrap the entire outside of exposed foundation in a cozy blankie would be a substantial piece of work if it was worth doing at all.

While the patch was too small to have any notable effect on the overall house energy demand, it survived its test in the outdoors just fine.  Nothing came along to molest the fiberglass and it didn't get soggy, just a little dirty on the surface which the wall does anyway. 

External insulation concept Anyway, now I knew what I needed to do if this was to move ahead.  I wasn't about to try digging all the way down to the foundation footing, that would just be crazy stuff.  But a modest trench as deep as I could practically make without destroying the yard would allow significant penetration into where the ground perhaps ran a little closer to seasonal earth temperature, and that might allow the wall to stay warmer in general and just help the big complex thermal picture.  I had a pretty clear vision of how this would look -- use the big version of the picture for the detailed section concept.  The Roxul sheets come in 4 x 3 foot sizes, and I figured that if I could use them in "landscape mode" and go the full 3 feet down from the top of the wall, that would be about the right balance between insulation value and work factor.  The Roxul would be well protected from bulk water by the metal overhang and the eaves up at the roofline, but if any water intrusion did happen it would need an easy path to percolate through and out the bottom if needed.  So the concept included a small stone-filled drainage channel at the outer bottom corner to carry water away instead of letting it get trapped down there by a tight surround of dirt.

  As I became more confident that the scheme had merit, I took a few measurements around the house and started a plan sheet, if for nothing else than to determine how much material I'd need.  I could see immediately that there would be quite a few fiddly-bits and places where partial pieces would be needed, so I arrived at a generous estimate.  I also stopped by the contractor desk at my usual local big-box to chat about sourcing the Roxul rigid board, and the guys there assured me that they could get it in convenient 3-packs through a supply-line called Primesource and even gave me an SKU number.  Roxul mentions this partnership on their where to buy page, so I squirreled that information away for when I figured I might actually want to act on it.

Which, cutting somewhat to the chase, finally happened about a year after the patch went on.  It was actually not an easy decision, because it was already late in September and I wondered how much good weather I'd have left in between other projects I had to work on besides.  It felt like my window of opportunity for that year was already starting to close, but I also allowed the possibility that maybe I'd only get half the job done before winter and then I could at least do insulated wall A vs. exposed wall B comparisons or something.  But then that would slow down the whole process of re-determining whole-house heat-loss figures after the job was fully completed.  Argh, what to decide?? 

One morning I decided to just stop being lame and wishy-washy and just *go for it*, and do as much as I could when I could and see how it would go.

My adhesion pull-tests had shown that the CIS type board held a slight edge in strength over the IS by being somewhat denser, at a slight disadvantage in R-value.  But I figured I needed all the holding power I could get, wood-hardener or not, and decided that the CIS was the most suited for the job.  And that's when the sourcing nightmare began.  When I eventually contacted the big-box back to price up some quantities, they seemed completely unable to find what product I was talking about.  After going around and around with them and Primesource directly it finally emerged that Comfortboard CIS is considered a "commercial" product and thus Primesource not only didn't carry it, they said that they were not *allowed* to carry it -- per corporate policy of Roxul themselves.  WTF??  More digging around finally got me in contact with a commercial supplier who could get me CIS, but I'd have to buy an entire pallet which was close to twice what I'd need and came in inconvenient piece sizes.  They quoted me for both IS and CIS, which at least gave me a solid fix on the price of two-inch material per square foot, which is a little over a dollar per for either type.  But suddenly CIS seemed to be out of reach for my small-quantity needs, because of stupid restrictive channel policies from Roxul Canada [which, incidentally, they deny the existence of].  With some reluctance I finally gave up and decided that IS would probably work just as well, and basically started over.

Which led to another nightmare as the disparate computers at Big-Box and Primesource couldn't seem to agree on per-pack pricing for IS, and they fully intended to charge me about *three times* what the stuff is worth.  Probably some confusion over units of "pieces" or "bags" but I couldn't convince anybody in the chain that they might have a mistake somewhere.  Back to the commercial supplier, then, who said they'd be happy to supply me ten packs of IS at the correct pricing and deliver it right to my driveway for an extra thirty bucks. 

Done!!  And then I fired a major rant into Green Building Advisor about this travesty, to warn other seekers of Roxul product that they might be in for an equivalent amount of, uh, administrative fun in trying to source it.

But now I was committed to doing this, and had quite a bit of work ahead even before any goods would hit the doorstep.  The supplier told me about a two-week lead time at the outside, and I wanted to prepare efficiently and be ready for delivery.

Starting a big brushing job The exposed wall had accumulated years of backsplash dirt and moss, especially on the north side, and would have to all be cleaned down to bare concrete for any hope of adhesive actually sticking to it.  I brushed down a quick test spot to see how well this would work and how clean I could get it, using a sequence of brushes from rough to fine.  It took a good bit of arm with the steel-wire brush to bust all the moss off of there, and raised a big dust cloud that indicated that this would definitely be a goggles-and-face-mask operation.

Cleaned up to front stoop While the wall-brushing task was a bit arduous it went reasonably quickly, not taking too long to get the front nominally done and then on to the rest of the house which was a little easier because it was less mossy.  I could definitely get all the way down to clean concrete everywhere, and thus have a good substrate to glue onto.  I dug a few inches down along the walls just to get it clean down a little lower, especially in the areas with the higher grade, as that bit of block would be part of the glued area too.

Concrete footer behind side stoop After taking all the fill dirt out from behind the side stoop and blowing the slot clean I finally discovered that there was indeed some sort of large concrete footer for the precast stair unit to sit on, that in turn had been poured right against the wall.  This was one reason the stoop needed to be re-leveled after being shifted out during the retrofit -- the pitch of this footer was all wrong in the first place, and my modified support with the bricks and all had clearly brought it much higher at this corner to correct for that.  I wouldn't be able to dig down past this thing, so I'd have to do the best I could with regard to insulating through the area.

Grand plan Along with cleaning the walls I took more measurements and tentatively marked the 4-foot intervals, to determine what sorts of cutting would be needed to work around features of the foundation, and updated my notes sheet.  This is a shot of it near the *end* of the job, after many of the little complexities had been dealt with.

With the wall's mortar joints now more visible I also took the opportunity for another round of concrete-caulking, to seal up any remaining obvious cracks and air-leak points.  A few new ones were discovered just below grade level, which given the permeability of the super-dry soil could leak just as much air as a crack above ground.

    Whee!   Playing in the dirt again

Spraddling the dig With the Roxul still somewhere in transit, I began to dig my trenches.  Any removed dirt would be contained on top of tarps, especially by the driveway.  As I worked on the first section I found that I should have possibly folded this tarp to be a bit larger, as the mound just barely fit on it.  Shoveling needed to be done carefully but was relatively easy, and for pulling rocks and roots and doing fine adjustments I had to take my little "digging stick" and get down in there but avoid caving in the edge of the trench.  Working under the existing house overhang made it all quite a bit more interesting in terms of balance and where I could actually brace myself.

Three feet down On the north and east sides of the house where the grade is highest, exposing three full feet of wall involved a respectable amount of excavation.  And that had to be three feet down from the lowest point of the siding and flashing metal, not just three feet of block wall height, since the pieces of insulation would have to be ducked under that outer overhanging stuff before getting attached.  So basically about five courses of block had to be visible, requiring my trench to be about 18 inches down along here.

I could almost wish I had the same apparatus that a nicely relevant "excavationless" NREL project had available, but that configuration wouldn't be vapor-open at all.

Corner dug out Rubble under the front stoop
While still waiting for delivery, I wanted to prepare as much as I could to start installing the moment it got here.  I laid down another ground tarp and started along the north wall as well.  As things advanced even farther I started scrounging up any available membrane I could find to lay dirt on, and kept going all the way to where the run would turn the corner onto the front stoop.  As I exposed the stoop's underpinnings, I found quite a bit of of air underneath and it was obvious that the stoop had been built in a very typical fashion -- the builders slapped together a form, dumped in whatever random rubble they had laying around the property and then poured lots of concrete over it.  I briefly toyed with the idea of trying to fill that space with sprayfoam....

Note how I brushed off a foot-plus of moss on the side of the stoop, but stopped there.  I intended to turn the inside corner and continue the Roxul out from the main wall a little ways, and apply that "flange" configuration anywhere else I couldn't insulate the main wall directly such as the other stoop and the basement bulkhead.  Concrete has lousy R-value but it's not zero, so by covering sort of a "tunnel" over some of what I couldn't directly thermal-break I'd still get a little insulation value from that.  Same principle as the long tunnel entrance to an igloo.

Delivery truck! The two-week lead time came and in fact had passed, I was already dug out from stoop to stoop, and the supplier wasn't responding to any of my status-inquiry email.  I was starting to feel a bit of time pressure ... like I'd have less and less nice weather to work in, and getting a bit anxious about just how much of this monster I'd actually be able to accomplish.  I finally timed out and called them one morning just to make sure they hadn't forgotten about me somehow, and was met by the happy answer "delivery is scheduled for *today*".  Awesome!  True to their word, a couple of hours later the truck arrived with my goods. 

My stack of Roxul! I helped the driver unload the stack from his pallet-jack since the stuff wasn't on a pallet anyway, and with the open trenches and mounds of dirt visible right along the driveway, he seemed rather interested in how what he'd just brought me was going to play into the project.

So I finally had my material: ten mostly-undamaged bags of 2" Comfortboard IS.  Now the real, uh, fun could begin -- when all those unknowns would start getting resolved.  Would this crazy thing actually work??  Had I bitten off way more than I could chew?  Didn't matter at the moment; delivery at long last had renewed my vigor, and it was a beautiful day for working on stuff like this, so I tore right into those packs and got started.

    Continue to Part 2: Installation

_H*   151030