Rooftop deck layer discussion. Can I make it thinner? Condensation issues? Other issues?
Greetings all! I’m planning a build with a small roof deck, and vertical space is at a premium due to height restrictions. Every mm counts. I also want to get R50-60 between the roof deck and the living room below. I want to do as much of the work myself as possible, and keep it as green as possible while also keeping it as close to zero maintenance as I can. So I’m open to suggestions on materials but I’ve already done a lot of research on the most efficient materials for this purpose and I’m pretty satisfied. If a material you want to suggest is not as green, durable, or as thin as what I list then please skip it. Polyiso, spray foam, XPS, cellulose, wood fiber, and fiberglass insulations are all out. If you have any other suggestions, or if my assembly will cause condensation issues, please let me know. Location is southern Colorado, elevation 6000 ft, climate zone 6a, 20″ of rain/yr avg. deck size is 11ftx15ft, over a “flat roof” with 1/4″/ft slope in the 11 ft direction. Through-wall parapet drains flush with lowest edge surface. I’m ok with spending money to do it right once since I plan to live there for 30 years (fingers crossed). The assembly needs to be as thin as possible.
Here’s what I have in mind, layers from the top to the bottom: 10-15mm thick tiles as the deck surface, plastic (PE) low-profile leveling pedestals starting at 1/2″ thickness and sloping 1/4″ per foot to 3 1/4″ thickness, seamless 60mil EPDM, EPDM bonding layer, 1/2″ CDX plywood, 8 inches of HALO GPS Exterra or Subterra (depending on psf rating engineer thinks I need) with 4.8 R value per inch for R38.4 glued together and placed in alternating directions using either 1″ or 2″ thick sheets, then 1/2″ CDX ply, vapor barrier, then 6″ tall 12 or 14 gauge steel C-channel style joists 16″ OC with 6″ (vertical) rockwool comfortbatts fitted snugly between (R3ish per inch, for R18), then 1/2″ drywall, latex paint.
18.5″ total thickness, about R60, plenty strong for a roof deck (to be confirmed by engineer). No rafters required below the ceiling, or additional structure eating up my precious headroom.
Can I make it thinner? Any issues with condensation? Any other issues? 10″ SIP-style screws would go down from the upper plywood layer to the steel joists.
Thank you
GBA Detail Library
A collection of one thousand construction details organized by climate and house part
Replies
I doubt the engineer will be ok with 1/2” structural deck. I would not place a vapor barrier below your structural deck. Most roof membrane manufacturers have a a secondary waterproofing layer that gets adhered to structural deck. But you want the structural roof deck to dry to interior and so that vb layer is poorly placed. If you must, install a smart membrane directly above Sheetrock.
Scuppers are ok but must be installed very well, don’t skimp on roofing installer. A better system for low slope on longer roof decks is an internal drain, as the natural loading of your joists bends toward the center of span. Another option would be no parapet and just drain off drip edge
I would probably stick with wood, I dont know what benefit steel joist give you and negate much of the cavity insulation.
The benefit of 6" steel C joists is they can span 11 feet in 12 or 14 gauge. Wood 2x6 can't span 11 feet. Also, according to Rockwool's 'effective R-value calculator', a roof assembly with steel C-joists has a higher effective R value than one with wood joists. My understanding is that's because the insulation is only separated by a very thin layer of metal web, instead of the entire width of a 2x6 every 16" OC. Check the 16.25" width of comfortbatt 291372 compared with the product they make for wood studs, which is 15.25" wide for 16" OC. Yes I could use a 2x10 or 2x12 wood joist, but I'm trying to keep the assembly as thin as possible- that's the main challenge.
Thanks for your comment on the vapor barrier location. I was thinking mainly about keeping interior humid air from going up, but I get what you are saying about the roof deck. Also I will check on roof drains but pretty sure that will either increase the assembly thickness, or lower the R-value, or both. Draining off a drip edge is an interesting idea, and is less likely to cause trouble. Finally, yes I will check with the engineer about the plywood thickness. That rigid foam board is so hard (very high PSF) that my main reason for having a plywood layer was just having something to screw down, but I will definitely ask.
You should look through the Carlisle details library to get a sense of best practice with flat roofs. They are very extensive.
Lvl are a good option for longer span thin joist
Nice- The Carlisle info looks great. I'll check LVLs again. For some reason nobody lists 6" LVL joist span tables- the few I've seen only have a 10-20% advantage over wood. Metal can be 100-300% better than wood for spans. Also nobody seems to make a 6" I joist. The ultimate thin floor is a metal sheet/concrete composite, which can be 3.5" thick and span 11 ft, but installation is a huge headache.
I recently did a project that the engineer was able to replace a 2x10 with 1.75x5,5 lvl on a roof deck. I dont know what your load is but I’m sure the engineer will let you know.
Measure_Twice,
Something is not right. Metal studs in an insulated assembly diminishes the R-value considerably more than wood ones:
https://www.buildingenclosureonline.com/blogs/14-the-be-blog/post/86806-effective-insulation-r-values-in-steel-vs-wood-framing
The difference between the Rockwool effective R value calculator and the Ashrae study is presumably that, with the Rockwool calcs, they are using the Rockwool product designed for steel C-channel joists, which are 16.25" wide batts (compared with their batts for wood joists, which are 15.25" wide). A C-channel joist can have the batt inside of it, because it's C shaped. However, in the Ashrae study, they probably used the same amount of insulation for both the steel and the wood, in order to have a control, and not put a variable into the results. No self-respecting scientist would use different amounts of insulation in a comparison experiment. Therefore, the C-channel was likely empty in the Ashrae test, which caused a huge thermal bridge.
No,
https://cfsei.memberclicks.net/assets/researchreport/RP18-1.pdf
"No air gaps were modelled between insulation and other components. The insulation was assumed tight to the gypsum and fit perfectly within the studs and tracks. This includes through the stud punchouts. Contact resistances were still applied. "
All metal stud assemblies use wider batts. Anything that was tested would have used the right sized (ie 16.25" wide) ones. Unfortunately the extra bit of insulation inside the stud space doesn't matter.
Not sure where the rockwool folks got their numbers from but all generic 6" 16OC steel stud assemblies work out around R7. With a full fill of spray foam that jumps up to a whopping R8.
I have used a while back DeltaStuds, essentially steel that is punched out for reduced thermal bridging which is better but still nowhere near wood.
I know it is hard to believe but that little bit of steel conducts a lot of heat.
If you want decent R value and thin, wood construction is your friend. For longer spans go for LVL joists.
Since you do have a lot of exterior rigid you can go with steel, as long as you are fine with the lower overall assembly R value.
For the rest as long as your above deck insulation is the same R value as bellow deck, you'll be fine in zone 6. I'm generally in favor of higher insulation but R60 done with expensive materials doesn't make much financial sense. ROI of that over code min (probably R42 assembly assuming U factor based compliance) is well past never.
This is a pretty simple calculator to work out your assembly R values. Also does steel studs in case you want to try it.
Even with cold weather de-rating, polyiso would still end up thinner and hold more weight.
+1 on skipping the vapor barrier. Not needed there and if anything it will make the assembly less robust as it prevents drying to the interior.
Measure_Twice,
https://buildingscience.com/documents/insights/bsi-005-a-bridge-too-far
https://www.greenbuildingadvisor.com/article/building-with-steel-framing
Thanks Akos- Yes I didn't have reason to think Rockwool would have an incentive to inflate steel framing stats, so I didn't think twice about their calculator- figured they are pretty scientifically-minded, but guess I was wrong. They must have just run the numbers based on there being additional rockwool in the assembly, without regard for the framing material- arg.
The good news is that 11 feet is just outside of wood's span capacity, and from what I can tell LVLs can go a bit past that and stay Ok in terms of deflection- so I'll check with an engineer but I'm pretty confident they will do the job.
Thanks for the info on the vapor barrier. I understand the concept of cold-surface condensation, but since it's so complex in the real world I'm happy to just go with an experienced opinion.
Steel is so highly conductive -- R-value indistinguishable from zero to three decimal places -- that a c-shaped piece effectively has the same conductivity as a solid rectangular piece of the same outside dimensions. You can't treat a 14-gauge c-joist as if it is 0.075" thick, you have to treat it like it's 1.5" thick.
Ok so that article is mixing data for that chart. The metal stud data is from the ASHRAE report, and the wood data is from somewhere else, so that's clear as mud. However I take your point: metal is conductive, and not ideal for an application where thermal bridging is a concern. I would love to know the basis of the Rockwool "effective R value calculator", whether it's merely theoretical, or based on experimental data.
Anyway, I should be able to make LVLs work, even if I have to sister 2 together. It's annoying nobody lists LVL span tables- Maybe it's because supposedly they are way stronger that wood, but not much better in terms of deflection- so technically their span is not much better when it comes to floors where deflection is critical.
Measure_Twice,
https://www.westfraser.com/sites/default/files/products/LVL/LVL%20Users%20Guide%20-%20Canada%20v0415.pdf
Thanks Malcolm, but notice the 5 1/2" LVL spans stop at 10 ft. That's the same as wood.
https://www.roseburg.com/wp-content/uploads/2021/03/Canada-Southern-Pine-1.75in-2.1E-True-LVL-PLF-Charts.pdf
Freyr Design thanks for the Roseburg chart- that's very helpful. I'll have to do some math to figure out something like a joist span table equivalent, but the data is key. Much appreciated
Ok Malcolm, you all have convinced me that the rockwool calculator is super wrong! Reality check appreciated. LVLs it is
Measure_Twice,
We've got a bit sidetracked by the discussion about what rafters to use. Going back to your original question, I agree with Freyr_design - for a small roof using a drip-edge to drain makes sense. I'd also consider reducing the R-value a bit as Akos suggested. You get a thinner assembly without much appreciable increase in heat loss.
I used pavers on plastic pedestals on this roof deck. It's a good system.
Ok good to know. So you agree mid 40s R value for roof in zone 6 is a good balance of cost/insulation?
Ha- I just ran the math, and I could run Doug Fir 2x4s with zero space between them. 3.5 inch deep joists... L360 Deflection, 50 pound load per sq ft. Let's see, 15 ft of them would be... 120 2x4s. Only $891 plus tax! :)
This is just for fun though. Cheaper to run 5.5" LVLs with insulation between them.