Dear Jamie

Jamie Wolfe, skilled contractor, devoted NESEA member and one of the most nimble minds I know, was curious about my previous post on frost-protected slabs.  He wrote the following:

I’d like to know more about the building envelope. I did some presumptive math (1/4 cord = 5MBTU/768SF=4557BTU/SF/YR) – pretty impressive.

So these questions:
What did you model the heat loss at?
Did you test ACH?
What about ventilation?
What’s the rest of the energy profile?

OK – I guess I’m asking for that second post!

Jamie, our little building isn’t all that impressive.  Actually, as modeled and so far this first season, it’s looking more like 5500 Btu/ft2-year, which is still pretty good.  NESEA members are achieving similar results in projects far more complicated than this one.  I’ve had the added advantage of working for myself: I only had to answer to myself and to the data, and by using my own labor, I could afford a premium building envelope.  But perhaps a quick sketch of this project would offer a few tips, so here goes.

 

The office has a simple plan and a simple form, keeping costs down and detailing simpler.  It lies on an east-west axis, the south face within 15 degrees of true south.  It’s heavily glazed to the south (8.33% of total floor area), lightly glazed to the east (2.6%) and west (2.6%), and no glazing on the north.  The layout is simple, but compared to we were working in before, the space seems luxurious to us.

 

 

 

 

 

 

 

Of course buildings last longer than we’ll likely be in practice.  So the design is adaptable, to make a small guest cottage or rental unit someday.

 

 

 

 

 

 

For simple structures like this, we use ENERGY-10 for energy modeling.  (We’re in the process of adding REM Design to our toolkit.)  ENERGY-10 begins with a base-case model and compares the effectiveness of various upgrade strategies.  For our base-case model, we assumed the following (all in Imperial units):

• Concrete slab foundation with R 7.5 beneath and R10 at the slab edge;
• 2 by 6 walls with R-19 insulation;
• 2 by 8 roof with R30 insulation;
• better-than-average air tightness (Effective Leakage Area about 37 square inches);
• Marvin “Integrity” windows with double-pane low-e w/ argon glazing, u-value 0.33, Solar Heat Gain Coefficient 0.30.

We modeled and explored a series of upgrades before settling on our first set of strategies:

• Under-slab insulation increased to R15 and slab edge insulation increased to about R25;
• optimum-value-engineered framing to reduce wood use and thermal bridging;
• a wall system with a whole-assembly R-value of about 33;
• a roof system with a whole-assembly R-value of about 45;
• a robust, continuous air barrier and other air-sealing measures to lower Effective Leakage Area to 9 square inches;
• energy-recovery Ventilation with an assumed efficiency (60%) and a schedule equalling 0.2 “natural” air changes per hour.
• efficient lighting, “Energy Star” equipment and other measures to reduce plug loads.

 

 

The model showed significant differences between these cases: a 60% reduction in heating loads (from 25,500 Btu/ft2-year to 10,200) and a 59.5% decrease in overall energy use.

 

 

 

 

I felt I could easily pay for and justify all these upgrades, especially by doing the work myself.  The next upgrade needed more thought: a significant increase in the window budget for superb Canadian units: “Thermotech” windows from Ontario, with insulated, pultruded-fiberglass frames and triple, low-e, argon glazing achieving the remarkable combination of a u-value of 0.16 and a Solar Heat Gain Coefficient of 0.42.  Marvin “Integrity” windows are an excellent product, the Thermotech upgrade was expensive, the return on investment was low even if we burned propane, even lower since we burn wood… but heck, this is what we do and believe in, and if we charged ourselves the full cost for fossil energy and climate disruption, it wouldn’t even be close.  We bought the Thermotech windows.

 

 

 

With the great windows, our model showed our energy use at 5500 Btu/ft2-year for space heating and 17,000 Btu/ft2-year for overall energy use.

 

 

 

But as Henry Gifford, Chris Benedict, Marc Rosenbaum and others remind us, it’s one thing to model low energy use and another thing to achieve it.  Here’s how I got dirt under my fingernails.  I fully expect that many of you will suggest smarter ways to do what I was attempting – that’s really the point of the NESEA blog, for us to put stuff out there in hopes of stimulating discussion and thought.

 

I made a “load-bearing Larsen truss.”  This is a ladder truss fully capable of bearing the structural loads imposed by this little building.  I made a jig from 3/4-inch plywood, allowing me to place two 2-by-3’s in parallel, 7 1/4 inches from edge to edge.  I used 1/2-inch OSB cleats top, bottom and at the third points, nailed with galvanized 1 1/4” roofing nails.  I made similar assemblies, with full OSB covering, for top and bottom plates.  I framed these “studs” 24 inches on center.  I made rough openings 1 inch larger than called for, then closed them in with 1/2” OSB.  There’s a 2 1/4” gap in these assemblies into which insulation can be installed, vastly lowering thermal bridging.  The framing members are straight and consistent, they use small sticks, and they took surprisingly little time to fabricate.

I wanted additional R-value, additional diminishment of thermal bridging, and a robust air barrier.  I decided that a relatively-thin structural insulated panel (SIP) would do the trick nicely.  It’s possible to buy SIPS with 1 to 2 inches of foam in them, plenty for my needs.  But I had a few concerns.  First, it was much less expensive for me just to buy sheathing and foam, and install my own “sandwich.”  Also, I was willing to forgo a little foam in areas where solid nailing is critical, such as bottom of wall, corners, and around window and door openings.  And we live in a very windy spot near the Maine coast, so I wanted adequate lateral bracing.  So I used a first layer of 7/16” OSB, nailed with 8d galvanized ring-shank nails.  I placed 2 by 3’s or 2 by 4’s on the flat at those critical areas.  Elsewhere, I placed extruded polystyrene foam, 1 1/2 inches thick, taking care to seal all seams and gaps.  Then I placed an outer layer of 1/2” “Advantech” sheathing, nailed with 16d galvanized nails.  This outer layer of sheathing wouldn’t be necessary if I were using a clapboard or plank type of siding, but I was using cedar shingles (FSC certified, grown in Maine and milled in Quebec), and I needed a nail-base for them.  Otherwise, vertical furring strips would be fine.

 

 

 

The roof is a short span, so 2 by 8’s, 2 feet on center, worked fine structurally.  I added a 1 1/2” by 1 1/2” strip of extruded foam to the underside of each rafter, then placed a 2×6 beneath that, held to the 2 by 8’s with OSB cleats.  I placed 1-inch “Thermax” foil-face, polyisocyanurate insulation in each rafter cavity, on 1-inch foam spacers glued to the underside of the roof sheathing.  This space provides a channel for roof venting that is probably unnecessary – unvented roofs work fine if they’re done right – but done anyway with an abundance of caution.  The air-gap and foil facing may also slightly reduce summertime heat gain from radiant heat transfer from the shingles.  I was very careful to seal the “Thermax” at the bottom of each roof cavity, to provide strong coupling to the foam on the walls, thus helping maintain continuity of the air barrier.  The rest of the rafter cavities was for dense-packed fiberglass, and there’s an R-7.5 insulator at each rafter assembly.

 

 

 

Like many others, I was trained to believe that dense-packed cellulose or fiberglass, installed properly in a well-built assembly, can be an air barrier.  I still think that’s possible, but harder to do well than we think.  I treated the air barrier and the thermal insulation as separate assemblies – except that my foam air barrier also helps reduce thermal bridging.  I also painstakingly sealed every hole, gap, penetration, etc., that I could find.  Fred Unger says that when it comes to buildings and water, we should “think like a drip.”  I’d add that when you think about air movement in a building, “think like smoke.”  Since the air barrier is separate from insulation, I tested its effectiveness before installing dense-packed fiberglass insulation.  The results were excellent.  The area of the building shell is 1762.67 square feet.  The volume within that barrier is 5256 cubic feet.  Our blower-door test, at 50 Pascals of pressure, averaged over 2 minutes, indicated leakage of 85 cubic feet per minute.  This translates to 0.97 air changes per hour at 50 Pascals, or 0.046 air changes per hour “natural.”  It’s an effective leakage area of 9 square inches – what we calculated in our model.  At 50 Pascals of pressure, the air leakage is 0.048 cubic feet per minute per square foot of building shell.  During the test, we located a few sources of leakage and fixed them.

 

(On a side note, I first reported these results to Marc Rosenbaum – to my horror, I had made an arithmetic mistake and grossly overestimated the building’s volume.  I’m usually careful with arithmetic, so a mistake like this is embarrassing, but when it happened with Marc, one of my heroes and a paragon of careful number-crunching, it was mortifying.  New frontiers in “geek shame.”)

 

So we had excellent air-tightness before most of the insulation, but we still needed the insulation.  The great folks from R.H. Price Insulation, of Montville, Maine showed up and did an excellent job filling all the wall and roof cavities with dense-packed fiberglass.  (Cellulose would be another great choice, but Bob Price’s crew does fiberglass, and in this kind of application we haven’t seen any difference in performance.  Loose-fill in an attic is another story, where we think cellulose is superior.)  Bob’s crew installed a vapor retarder called “MemBrain.”  This is a polyamide plastic that changes permeability in response to difference in vapor pressure between an interior space and an insulated cavity – in other words, it can block vapor transfer into the cavity, or allow drying to the interior by vapor transfer out of the cavity.

 

 

 

With a building this tight, it was obvious that we would need mechanical ventilation.  I bought and installed a “RenewAire” energy-recovery ventilator.  We like these units: they recover sensible heat and humidity, they don’t need a condensate drain or a defrost cycle, and they are relatively efficient in the use of electricity.  We control the ERV with a programmable-schedule electrical outlet and a “run-time-percentage” controller.  I took great care with the installation, keeping duct runs short and well-sealed, and installing the best backdraft dampers I could find.

 

 

 

 

The ERV is half of the “mechanical system.”  The other half is a very small, well-crafted wood stove by “Mørso” of Denmark.  This is their model 1440.  It’s a convection stove: cast-iron baffles at the sides allow air flow there.  This keeps the sides cool enough to touch when the stove is firing, and lowers the side clearance needed to just 8 inches.  Unfortunately, it’s not available with a direct outside-air feed – I don’t know of any small wood stove on the market that does offer one.  I compensated by located a fresh-air supply from the ERV in the ceiling directly over the stove.  Even with that, it takes some care and experience to manage the wood burning properly.

 

 

 

 

 

 

 

Such a little stove requires little sticks of wood.  Our ENERGY10 model had indicated 5500 Btu/ft2-year annually for space heating, or a total of 4,224,000 Btu.  If you assume that the wood stove efficiency is about 65% then you’ll need about 6.5 million Btu worth of wood.  Seasoned mixed hardwood contains about 24,000,000 Btu, so our space heating needs are just over a quarter of a cord of mixed hardwood.  (This assumes that we’re keeping the upper level at the same temperature as the main floor, which we don’t, so our actual fuel need may be less.)  We need so little firewood, and it has to be in such small chunks, that I wondered if I could provide our fuel without cutting down any trees.

 

I could.  There are techniques called “coppicing” and “pollarding.”  Coppicing is cutting shoots of woody plants that sprout back.  Pollarding is cutting selected branches from trees that re-grow those branches.  In my case, I did the coppicing on Black Alders and the pollarding on Dutch Timber Willows that I planted years ago.  I’m so happy with the results that I took cuttings from my willows, sprouted them, and planted some more.

 

 

 

Alder contains 17.2 million Btu per cord, and willow only 14.5 million.  Since I mixed them roughly fifty-fifty, my fuel only contains an average of 15.85 million Btu per cord, 34 percent less than mixed hardwood.  With this fuel, we need at least 0.4 cord.  Therefore, I played it safe and cut about 1/2 cord of willow and alder.  I also bought a small supply of compressed-sawdust bricks called “Biobricks” as an insurance policy.  I calculate that these provide about 15,000 Btu each.  We’ve had a colder-than-average winter, but as of March 15, we’ve used about 80 “Biobricks,” or about 1.2 million Btu, plus about 1/4 cord of our “junky” firewood, or about 4 million Btu, for a total of 5.2 million Btu.  We rarely need to make fires now, with the moderating temperatures and strengthening sunlight.  It looks like we’ll meet or beat our calculated heating energy.

We’re delighted with our new little building.  We’re proud of its performance, but know that with hindsight we could have done a little better, that it’s much, much harder to pull off this kind of performance in more-complex projects, and that astute readers of this post will offer all kinds of comments and suggestions that we haven’t thought of.  Now we just need to start saving funds for an eventual grid-tied photovoltaic system to offset the 3 kilowatt-hours a day we use for electricity.

---