Coordinating the roof

Submitted by Ted on June 8, 2011

I was on the phone with Marc the other day and he pointed out something that hadn't really occurred to me until he mentioned it: when we hired Ben and Eli, we took on a new job. One of the key things that you will hear from anybody who has any experience building a passive house is that your team has to be on the same page about what you're trying to do, or important things will fall through the cracks, and compromises will be made that make a lot of sense from the viewpoint of one participant in the project, but create a huge problem from the perspective of another participant.

We've made a ton of changes to our house since we first started brainstorming about it with Marc a year and a half ago. A lot of them hit quite recently, when Ben talked us out of our Shogun foundation, and when Eli talked us into flipping the roof. We had to give up some things we'd been looking forward to, but we tried to keep a razor focus on the air barrier and insulation, because together those give us what we really want in this house: a consistently comfortable indoor environment.

But one point that Eli really pushed us hard on was the roof. In order to understand what the controversy was, I need to explain the original roof design that Marc came up with. On the way, let me point out that because Andrea and I are so interested in the details of this process, we never have a conversation with our honored teachers in which we just listen to what they tell us and do it. Consequently, nearly every aspect of the house has our fingerprints on it in some way, and we don't always remember which things we decided to do were "nice to have" and which were crucial. The roof design wound up being one of those things.

The problem with an almost-passive-house roof is that it has to perform consistently in two senses: first, the insulation has to be consistent. We can't have it settling down from the top, bunching up on the bottom. Way back when, when we got our first estimate from Keith at Thermal House on insulating the roof, we were absolutely floored: it was going to cost about twenty thousand dollars. Why so expensive? Because we'd chosen to do a cathedral ceiling, not a sensible flat ceiling.

Flat ceilings are easy to insulate--you can just dump the insulation on top of the ceiling joists and spread it out, or if you want to get fancy, use dense-pack cellulose. Because the ceiling is flat, there is nowhere for the insulation to go. You wind up with an irregularly-sized air gap above the ceiling, since your roof certainly isn't flat, but in most situations it should be pretty easy to get a nice thick blanket up there.

With a cathedral ceiling, you don't have the luxury of flat. Keith solved that problem by using a really expensive foam product that he was confident wasn't going to move. But we can't afford to spend $20k insulating our roof. So Keith and Marc and Andrea and I did quite a bit of negotiating. At the time, we'd been planning to use open-web trusses to support the roof. We were confident we could get trusses that would span the distance from the north to the south wall, and would support the dramatic 7' overhang we wanted on the north. But insulating open-web trusses is really hard, as Andrea explained in her recent post.

What we came up with was that instead of using trusses, we'd use I-joists. Andrea did some research on the Internet, and found a supplier for a 24" deep I-joist that would span the distance we wanted, and additionally would support the cantilevered overhang. Marc and Keith both liked these better, because Keith was confident that he could do a dense-pack cellulose installation without any voids using the I-joists. Marc liked the I-joists because they cause very little thermal bridging: the webbing between the flanges on the I-joist is only 7/16" thick, so even though its R-value is substantially less than that of the roof, there isn't enough of it to create a problem.

At this point I need to talk about roof venting. There are two kinds of roofs: warm and cold. A warm roof is a roof with no air vent. They are popular in applications like ours, because building roof vents is fussy when you can't just vent the eaves into the attic. Typically a warm roof consists of a ceiling, rafters of some sort with insulation packed between, a layer of waterproof sheathing, a layer of insulating board (typically polyisocyanurate), a waterproof layer, and then the roofing material. So it's not a particularly simple roof.

So we got this brainstorm to do a vented roof, even though we're doing a cathedral ceiling. Marc had a clever design. He proposed that we cut a piece of insulation board to the spacing between the webbing of each I-joist. We could then cram this up against the top flanges of two I-joists. This would create an air space between the insulation board and the sheathing that's nailed to the top of the I-joist. Now we have a clear vent running the entire rise of the roof. It's pretty easy to construct, and it'll keep the roof cold.

So that's the roof design we'd been planning on for nearly a year. But then Eli comes along, and he has a number of problems with it. And they are real, serious problems. First, the top flange of the I-joist is going to be really cold on a cold winter's night. The bottom flange is going to be really warm. So the bottom flange is going to expand, and the top flange is going to contract. The roof is supported on both ends, so it might bow in the middle, or the I-joists might warp a bit, but the bottom line is that there's going to be some unwanted movement. Eli felt that this was potentially a big deal; Marc wasn't sure it was, but he didn't claim it definitely wasn't, either. I have no informed opinion on the matter—I have to trust Eli and Marc.

The bottom line is that this got Eli to thinking about some other advantages of trusses over I-joists. One of the big problems we'd left somewhat unsolved with the I-joists was how to construct the overhangs. Remember that there are four overhangs. We'd been planning to stick the I-joists out over the north and south overhangs, and didn't really have a solution for the east and west overhangs. Eli didn't like the idea of sticking the I-joists out like that, because it was going to potentially expose the webbing in the I-joists to moisture on the ends, and possibly wick moisture into the roof. He wasn't entirely clear on how big a problem this was, but it was definitely a concern.

Also, whereas the I-joists are only strong if the webbing and the flanges are preserved, trusses can be engineered so that what sticks out past the envelope is thinner, and yet still provides enough support. We'd been trying to figure out how to avoid having the overhangs be two feet thick, and were talking about cutting off the bottom flange and attaching something to the webbing higher up. Eli didn't like that idea from an engineering perspective.

A final advantage of trusses is that they are open, which means you can stick things into them. This gave us a way to do the rake overhangs—the overhangs on the east and west sides of the house. With the I-joists, we had no answer for the rake overhang other than nailing a box to the side of the roof. This could probably have been made to work, but making it structurally sound would have been quite involved.

I heard about Eli's counter-proposal on roof structure while we were up at the site seeing what had been done, I think on Monday. We'd been talking about using trusses for a while, but on Monday we came to the conclusion that we really couldn't use the I-joist roof as originally designed, and we were starting to go down the path of figuring out how to insulate the trusses. We also started thinking about going back to a warm roof, because the cold roof was starting to look like it would be pretty difficult to build with trusses.

I should mention that in the background we'd been hearing about another scheme that Ben, the structural engineer, was thinking about, that involved a two-layer roof. I didn't really like the sound of that, because it sounded like a lot more work, so I hadn't made the effort at this point to find out precisely what he was talking about. I had visions of having to insulate two sets of cavities. Not a happy thought.

On Tuesday, Marc called Andrea to weigh in on the rash of changes we were suddenly making. I was sitting in my chair minding my own business, working on something or other, and suddenly Andrea thrust the phone at me and said, "You talk to Marc." So I wound up explaining the whole truss thing to Marc, and Marc of course was asking his usual intelligent questions and making comments, and I was starting to feel pretty sheepish about where we'd gotten with Eli.

I mentioned some harebrained schemes for mitigating the problems with trusses, and Marc was pretty patient with me, but he reminded me of what we'd been trying to accomplish with the trusses, and the complexity of getting a good air barrier with trusses, and so forth, and by the end of the conversation I was back to believing that I-joists were the right thing, and that we just had to figure out how to make them work. Marc also explained why warm roofs have insulation on top. In a warm roof, you have warm, moist air below, a piece of wood, and then a moisture/vapor barrier. This means moisture can condense on the wood, and has nowhere to go. Venting the roof provides a place for the moisture to go. Without it, the wood could rot.

Andrea reported the outcome of this conversation to Eli via email. A while later Eli called me and set up a Skype session so that he could draw stuff in his CAD package while I watched. What he drew was a roof that used I-joists for structure, but then had a venting system on top of it. The venting system is essentially a second roof, with ribbing that mostly runs parallel to the I-joists.

The eaves are cantilevered on the ribbing, but the ribbing is also used to form ventilation channels. The corner overhangs are done by installing a rib at each corner at a 45-degree angle to the I-joists. The rake overhangs are done by installing ribs at 90 degrees to the I-joists. Throughout the rake and corner overhangs, ribbing is also installed parallel to the I-joists to form vent channels, and vent holes are drilled in all the ribbing that is not parallel to the I-joists, so that even the corner and rake overhangs are still vented.

Of course, this roof has a problem as well. Because it's so nicely vented, if you get a wind across the roof, it can form a relative vacuum at one of the vents. This vacuum can then suck rain or snow into the vented part of the roof, which falls into that roof. If the roof isn't waterproof, the wood gets wet. So Eli is proposing to put down some kind of waterproofing.

One thing that just occurred to me while I was typing this, though, is that now we essentially have a warm roof with wind channels above it, because the bottom layer of sheathing has a moisture barrier on it. So I need to try to understand whether that aspect of this roof really makes sense. But at least from the perspective of supporting the eaves, this roof is the best design we've seen so far, and I'm feeling pretty good about being able to iron out the details.

But now perhaps you can see what an interesting job the prospective owner-builder faces when trying to get a passive house built with the help of a lot of really talented and experienced people. I feel very lucky, but Marc is right in saying that this is a significant job.

Here come the joists!

Submitted by Andrea on June 8, 2011

Today I sent this e-mail to our energy guru Marc:

The joists in question are 24-inch roof joists from Nordic Engineered Wood in Quebec. These bad boys will allow us to have a clear span roof (no internal bearing walls) that's stuffed with 24 inches of insulation (mostly if not entirely cellulose).

Open-web roof trussesThe reason Marc wins this round is that Eli (the builder/architect) and Ben (the structural engineer) were agitating for flat trusses. Trusses have some structural and workflow advantages, but joists have the virtue of being extremely easy to insulate.

For those of you not in the know, trusses are made of dimensional lumber (usually 2x4s or 2x6s) joined together with metal plates. They are exceedingly strong and relatively inexpensive. The problem is that it's hard to properly insulate all the gaps, and the wood and metal turns into a thermal bridge when it traverses the building envelope.

As an alternative, Marc has long wanted us to build the roof with I-joists. I-joists are an engineered product, which means they aren't made from old-fashioned wood like the Pilgrims used. Instead, they're an unholy adhesive-bound combination of solid wood flanges (the top and bottom bits) and OSB (particle board's stronger cousin). They're called I-joists because the cross-section resembles a capital I.

I-joistsUnlike trusses, I-joists are very easy to insulate, and the relative lack of material means less thermal bridging. Our I-joists will be a flabbergastingly-deep 24 inches, which is practically unheard of for residential construction. We want this depth for insulation and also for strength.

It was initially hard to find joists this deep. In residential construction they usually max out at 16 inches, and although Weyerhaueser's commercial line includes 24" joists, they aren't available in New England. Fortunately I discovered that our French-speaking neighbors to the north make deeper joists that are relatively easy to obtain.

Ted promises to write a post explaining the technical challenges with using I-joists and why Eli and Ben weren't initially on board.