I have a client who needs your help. In the early 80's, he and his wife built a beautiful off-grid home on a small island near Fort Myers, FL. It currently serves as a convenient getaway, but as they approach retirement, they want to install a limited amount of air conditioning in the form of a couple of small, high-SEER ductless mini-splits.
The aging 12V PV system is quite small -- the refrigerator is propane powered, and there's no well (fresh water supplied by rainwater collection). So he's going to need to upgrade the power system to support the mini-splits. To keep costs reasonable, some envelope improvements are warranted.
It turns out, the gorilla in the room is the uninsulated metal roof. It's supported by beautiful site-built beam trusses and the exposed roof panels serve as a decorative ceiling. You can only imagine how hot that surface gets, making the home virtually uninhabitable for 4 or 5 months a year.
I'm not sure what's the best approach to insulate this roof. The owner wants to avoid pulling up the existing roof, which is seriously bolted down, successfully weathering Hurricanes Wilma & Charley and numerous tropical storms. Moreover, the owners have a strong preference for preserving the aesthetics of the exposed metal interior, which means working from above.
I'm thinking 4" of XPS or poly-iso covered by another metal over-roof. Keep in mind the roof serves as primary collector for fresh water so shingles are a no-go (aside from blow-off risk from the inevitable storms). However, I don't know enough about this to advise on attachment and edge details. Nor do I have a feel for cost, other than it seems obvious that working from below would be less expensive since it wouldn't be structural and he could use a less expensive insulation product.
I've attached several images for reference. I know we have some savvy retrofit gurus among the membership. I'd appreciate your advice!
@Stuart, I wasn't considering the copula windows... yes, those would indeed need to be re-worked to make room for a build-up. Thanks for catching that.
Only because of the great advice given by everyone who posted here! I've asked the owner to weigh in.
The cupola windows are the only place there would be a problem if 4" of foam were added above the existing roof. In an earlier response I noted that I would probably have to replace those windows, which isn’t a huge task.
I had looked into gas air conditioning, given my positive experience using gas refrigeration since the mid-1970s. We haul 20-pound aluminum propane tanks now to run our stove, tankless water heater, and a small refrigerator/freezer, which hasn’t been a big deal. (I hauled three tanks last week by kayak!) But without a natural gas pipeline or any way for a truck to deliver us propane, gas air conditioning doesn’t seem feasible. I hadn’t even considered how much electricity it might draw....
I hadn’t ever considered insulating internally in the space between the beams. The biggest advantage, as you point out, would be the ability to do this only in the room(s) being air conditioned, to minimize the aesthetic impacts on the rest of the house (not to mention how difficult this would be). BUT, we wouldn’t achieve any MRT reductions in the rest of the house as we would if we put insulation on top of the entire house. So I’m inclined against considering this further, except as a potential retrofit to improve AC performance in a critical room.
I agree that it would be difficult to insulate from the exterior. It does look like at least a sheet of Polyiso (PIC) could fit between the purlins and over the beams. Additional layers would likely cover the purlins. I cannot tell from the photo how deep the purlins are. The PIC needs to be covered with something for at least aesthetics. Assuming the owner is aware of the dangers of PIC in a fire and decides against a sheetrock thermal barrier (assuming there is no code enforcement), then corrugated roofing would be an appropriate covering, just be sure to seal at least the ends so that wind cannot get into the channels?
FYI, those screws holding the existing roofing typically use a rubber washer. That rubber will deteriorate and fall apart, leading to roof leaks. They should be inspected periodically and replaced as needed.
I am with Brad Cook here on all points. But will concentrate on the rubber grommeted screws on the roof. Deterioration is on an exponential curve, and, as the rubber grommets go, the roofing metal is less damped, and more vibration allows the holes in the metal to enlarge and create bigger problems to solve.
However, this place was not built yesterday, and so the owner being a carpenter probably has these fasteners well in his maintenance purview.
Exposed rubber washers would disappear in a hurry! I used a screw assembly with a stainless cup washer that completely covers the rubber washer that compresses against the roof. The purlins are very dense; I’ve had very few of these screws come loose over 35 years -- except for a couple of dozen that weren’t tight after our close encounter with Hurricane Charley, and those were mainly on the hip flashings which took quite a beating.
The purlins are 1 3/4" thick.
I don’t know about the fire dangers of PIC or other foams. I worry plenty about fire, given the remote location, plus all the heart pine I used to build the house. The house is fully sprinklered but I don’t know if that provides much protection from fumes from burning foam.
This brings up a larger question for me – would the fire danger from foam insulation be greater if the foam were below the existing metal roof, instead of above it? Or would the danger be about equal?
Bill, you just posted a number of responses in different areas of the thread. I will try to respond to all of them here.
Any foam based insulation has a lower ignition point and they all give off dense poisonous smoke. The codes in the US usually require a minimum of a 15 min. thermal barrier over such insulation, which can be 1/2" sheetrock. The foam is much less of a danger on top of the roof.
It sounds as if you are leaning strongly toward insulating the top of the roof. Although more difficult and more expensive, I would agree. Keep in mind the 30 year rule: "Major renovations are typically done about every 30 years", so plan accordingly.
If you can deal with the cupola windows and the roof penetrations (and we only see part of the entire roof in that one photo), I would add 4" polyiso, such as Dow Tuff-R, with staggered joints, taped with foil tape.I would cover that with 5/8" or 3/4" plywood and use 7" FastenMaster's HeadLOC screws to screw the plywood to the purlins. Then screw a new, reflective or very light colored roofing onto the plywood. I would also use black (UV resistant) one-part foam to seal the corrugated ends of the roofing, so that wind cannot get under it to pull it up. Cover the edges of the insulation with 1" Boreal or similar trim. You could use Todol's PurStik one-part foam adhesive as well as fasteners on the trim.
At the cuploa windows, taper the PIC down a foot or two out.
IR non-contact thermometers use infrared to measure temperature. They use assume that the surface is reasonably close to a perfect black body, in that the surface allows little infrared to reflect back or pass through. It assumes that almost all of the IR read by the thermometer is given off by the surface and not reflected or passing through from another source. Your ceiling is white and somewhat reflective on the inside, so your thermometer is probably not accurate.
When the roof is hot, it will radiate a lot of IR to the interior. If you are standing in line of sight of the ceiling, you will feel that IR. It is similar to stepping outside into the sun. It will feel a lot hotter standing in the sunshine.
BTW, for SPF (spray polyurethane foam) you can buy kits of up to 600 board-feet that can be dispensed right from the package(s), and can be brought in a small boat, but it is expensive. Professional applicators use 55 gal. drums and special compressors and special pumping equipment that is not conducive to a boat ride. It is also difficult to spray any foam very evenly. A good professional applicator, using a truck mounted rig, can spray within about +/- 1/2". that will still leave gaps where wind can get under the roofing.
I may end up working from inside, a room or two at a time -- but my preference would be to do the entire house at once, which as a practical matter would mean putting a new roof on top of the existing one. This may end up being cost-prohibitive, but maybe not.
Foam boards would provide a dead flat surface and avoid inevitable air pockets between spray foam and the new final roof.
Given the location of this house on an island without road access, the plywood layer you describe strikes me as the biggest stumbling block – not in cost, but in logistics. Every board and every fastener comes over in my small boat, then gets hauled along narrow paths through the woods to the house. I used plenty of plywood to build this house, but I was 35 years younger and had a surplus of time and muscles. I plainly recall hauling sheets of 3/4" plywood through the woods, one at a time, then hoisting each one up 8' to become the elevated subfloor. Repeating the haul job and then lifting each sheet all the way to the roof is a daunting prospect; and having the new roof surface attached to plywood instead of directly to the purlins might be a step backwards in hurricane resistance.
By the way, I used 3/8" plywood on both sides of most walls partly because it was so much lighter than sheetrock (and would survive water damage). Although hauling that was a manageable task, 3/8" plywood wouldn’t suffice to hold a roof down in this climate.
So my question is: what is the role of the plywood layer in the system you describe? How essential is it?
Somewhere I recall hearing about roof panels that have a thick foam core encased in metal roofing top and bottom. If something like that were available to me, especially in long pieces that would run all the way from the cupola to the eaves, could I skip the plywood layer? I guess a hurdle would be finding fastenings long enough to go through these roof panels directly into the purlins – something like the HeadLOC screws you suggested, but driven through a rubber washer covered by a stainless steel cap.
I believe that it would be difficult to screw a 5-6" screw through those layers into a purlin, and those screws would not only have to be the right length, but would need a washer for every screw to keep the roof watertight. The plywood would require fewer fasteners and would not need to be waterproof. The plywood would allow for using conventional roofing screws.
As an alternative, you could use 2x3s or 2x4s screwed to the purlins with HeadLOCs, and fill the space between them with 1 1/2" PIC.
PIC can be compressed or dimpled if you over tighten the screws through the roofing, especially with the cheaper PIC with 20 psi compression strength.
Responding here to my earlier comment: "Somewhere I recall hearing about roof panels that have a thick foam core encased in metal roofing top and bottom. If something like that were available to me, especially in long pieces that would run all the way from the cupola to the eaves, could I skip the plywood layer?"
I just found an example (unfortunately in steel instead of aluminum): the roofing company MBCI makes standing seam insulated roof panels with painted steel (top and bottom) around a polyurethane core that's 2" to 6" thick.
If I put the panels like these on top of the existing roof, I could fill the gaps between the existing twin ridges with strips of thin foam.
I have no idea how the concealed fastening system for these panels would work. I also haven't checked to see how heavy full-length panels would be -- maybe the logistics would be worse than plywood!
Brad Cook wrote: "Your ceiling is white and somewhat reflective on the inside, so your thermometer is probably not accurate."
How could I test this?
If I put a piece of blue masking tape on the ceiling and gave it a little time, would it reach the ceiling's true temperature and let me use the IR thermometer to compare its reading with the surrounding ceiling?
Yes, a piece of blue or black tape would have an emissivity value much closer to the ideal number of 1. Be sure that the area is large enough to read. If I recall, those IR thermometers read an area that gets wider as you move away from a surface, and the red laser is not pointing to the exact spot that it reads. Read the instruction manual for the thermometer.