Floor Insulation and Moisture Concerns in Hot Climates, Vapor Barrier Sandwich
In a house with a *fully* open crawlspace in a hot/humid climate, is it safe to insulate a plywood subfloor from below – even when the finish floor is impermeable (laminate, most tile, some hardwood ULs)? The dreaded “vapor barrier sandwich”…
I know that plywood is a vapor retarder… but if it has no way to dry out (because of foil-faced rigid foam below it and, for example, an impermeable tile underlayment like Ditra above), what happens to the moisture it takes in? “What moisture?” Well, s*** happens… maybe warm, humid summer air finds its way into the joist cavities. Maybe a big spill in a kitchen or bathroom makes its way to the subfloor (boards separate, grout wears out). Maybe someone leaves an AC setpoint too low.
Here’s Dana and Martin discussing the matter (Replies #3-5, 18)… Dana seems to share the concern, Martin not so much, but I wonder how they’d feel about a hot climate? Reply #14 is scary. https://www.greenbuildingadvisor.com/article/insulating-a-wood-framed-floor-assembly
Also, could a too-thin insulation layer cause problems in a warm climate? Dana once wrote: “In an air-conditioned house the bottom edges of 2 x 8 or shallower joists can drop below the dew point of the crawlspace air unless you add an inch of unfaced EPS.” It’s one thing to say the joists are thermal bridges (if the joist bays are spray-foamed), code only requires R-13 floors in Zones 1-2 – and 7 ¼” of lumber (9.06) + ¾” subfloor (0.94) + ¾” hardwood (0.68) + air film (0.92) is nearly R-12 by itself! With 2×10 joists, these “thermal bridges” would be *above* code-mandated R-values. Am I missing something?
EDIT FOR POSTERITY: I’ve since learned that I was missing something obvious — if only, say, 2″ of a 7 1/4″ D joist is covered with foam, then the thermal bridging does not start at the bottom edge of the joist (for that whole R-12), but rather at the point where the foam ends… so, you only get 2″ of the R-value of lumber, not 7 1/4″. So, for each time I run the numbers for a given foam thickness, I’ll need to enter a different u-value for the amount of joist that I’m leaving exposed.
GBA Detail Library
A collection of one thousand construction details organized by climate and house part
Replies
I recommend this article:
https://www.buildingscience.com/documents/insights/bsi-009-new-light-in-crawlspaces
In an insulated assembly you always want a vapor barrier on the warm, humid side. Lstiburek in that article recommends foil-faced insulation below the joists, with the seams taped.
I’ve been planning to follow Lstiburek's recommendation, which is why I have the above concern. He harps on vinyl, but the fact is that ~no drying will happen through (for example) the majority of tile installations either (nor, of course, through foil-faced polyiso). Putting vapor barriers on both sides of sheathing is widely considered bad practice -- but that's what he is advocating for subfloors in crawlspaces...
This is a dilemma you run into with basements and unvented roofs too. The vapor drive is from hot to cold. What do you do when the cold side can't dry? You put a vapor barrier on the warm side, and enough impervious insulation on the cool side that the dew point lands within the impervious.
Also note that because of stack effect, vapor barrier on the wrong side (WSVB) walls need less cold side insulation than WSVB roofs. Logically, a WSVB floor needs even less (in cold weather). In hot weather, stack effect reverses - if there is air movement, it's likely to be pushing cool, dry air into the floor assembly. Summary: exposed floors are lower risk that similar walls/ceilings. But they still need proper construction.
Total R value (ie. 2x8s vs 2x10s) has very little effect on moisture. And some time (vs substantial time) below the dew point isn't a problem.
Floors may be lower risk, but I've seen lots that were a total mess. Typical is a crawl space that lets in enough air to be at ambient humidity but not enough to be at ambient temperature. Joists are packed with fiberglass, no real air sealing or vapor barrier. This space is damp and cool year-round, in the summer you get condensation forming and no real drying in the winter. The insulation is wet, the joists are moldy and rotting, the floor above is buckling.
With an insulated floor you can get that situation, where the exterior is cooler and damper than the interior year-round, but you can also get the situation where the exterior is basically ambient air, so it's like an exterior wall turned sideways. And everything in between. You want foam against the cold side, and it's not always clear which side that is going to be.
"you always want a vapor barrier on the warm, humid side."
But according to Joe in that 'new light' article, the warm and the humid sides are not the same side (not always anyways).
He is assuming a situation where the house's interior temp around the floor is higher than the ground temp (which radiatively couples to the underside of the floor assembly). This won't always be the case, but apparently it can be. Yet the introduction of outdoor moisture leads to higher vapor pressures in the crawlspace (higher dewpoint) than inside the house. Bucking the 'vapor drive is from warm to cold' mantra. While that is almost always true, I'm weary that some folk start to think it's the actual governing principle meaning it cannot be bucked. The governing principle is vapor pressure.
I have to admit I have a hard time fully following his logic here though and it seems like it would only apply to a very narrow set of conditions.
He claims the joist is coldest—and therefore wettest—on the underside facing the ground due to coupling with ground temp. Ok makes sense for parts of the year in certain climates. Then he says that vapor drive is upwards, to the inside. Ok makes sense. And then, if there is a low permeability flooring, the vapor drive causes moisture to accumulate at that flooring causing moldy floors, etc. Ok. But in that case, will the RH still be higher at the bottom of the joist than at the top. The moldy and rotting floor would imply otherwise. If there is a problem at the floor sheathing, which is also a wood product that cares about RH, it seems a bit contradictory. At the least, it would seem the joist is susceptible to high equilibrium moisture content throughout the assembly. (speaking about fiberglass filled cavities, no ground side vapor retarder)
He then goes on the clarify that a very open crawlspace that allows enough air exchange to affect the thermal balance won't have the cold-bottom-of-joist issue, and in such a case, csspf can be used with exposed bottom sides of joists.
But in any case, he does recommend a vapor barrier on the bottom side of the floor sheathing some distance away. If going over the joists continuously, foil faced foam (foil facing the ground) being nice to help decouple the assembly from the ground temps.
Is there a winter problem with these assemblies?
I think it's counterproductive to mix analysis of "fully open" crawlspaces, where presumably the crawlspace approximately matches outdoor conditions with more sealed off crawlspaces (often quite humid and cooler).
A cool (eg 65F) crawlspace won't be creating a condensation risk at a 75F floor surface vapor barrier. But 85F and 80% beneath the floor will (dew point = 78F).
what does 'fully open' mean to James? I don't know.
I'm not sure what you mean by 'a cool crawlspace wont be creating a condensation risk.' Joe seems to think a 65 degree crawlspace could indeed cause condensation issues at the floor sheathing. Or I'm misunderstanding something with your statement or his.
Either way it doesn't really matter, the issue seems to be with double VBs and whether we should be comfortable with them in general. Unforseen wetting etc.
It's pretty clear what you want if you can make it perfect and not have spills, etc. It's less clear when those two criteria aren't guaranteed.
"Fully open" to me means something like what's shown in Photos 6 and 7 here (lattice is just cosmetic): https://www.buildingscience.com/documents/insights/bsi-009-new-light-in-crawlspaces... assuming the bottom edges of the joists are adequately high off the ground to mitigate the radiative coupling Lstiburek points out (WAG = ~3'+). I've worked on lots of houses like this, and it's a very forgiving assembly (assuming you're in a climate with low heating needs). I've also worked on lots of encapsulated crawls. I can tell you that, in SE Texas, the top 10% of the latter (workmanship and materials) perform amazingly, 70% are liabilities waiting to happen, and 20% should just say "Termites and Mold Only! No Vacancy!"
Getting back to the original question, an interior side vapor barrier with significant inward vapor drive has risk, even if you try to keep moisture out of the assembly (with an exterior side air sealing/vapor barrier). Options are the same as a cold climate roof - one is adding a vent space under the WSVB. Or add some cold side foam to warm the wood sheathing. Or let it dry to the cold side with something more vapor permeable (might not take much to get a flow-through perm ratio). Or make sure the cold side allows more airflow than the warm side (air "leaks" are often overlooked as a drying mechanism).
Your view makes sense to me. What does not make sense is Lstiburek's implication that you'll be fine as long as you avoid vinyl flooring. I think that "vapor permeable flooring" isn't all that common in practice. There is no appreciable drying happening through the vast majority of tile installations -- water or moisture might get through barely-detectable cracks in grout, but it's not necessarily "smart" enough to find its way out. And many hardwood underlayments are ~impermeable. And laminate? Point being, this issue is far more widespread than his focus on vinyl indicates -- and yet the recommendation seems to be, all over GBA and BSC, to put a vapor barrier under the subfloor (in the form of foil-faced foam). That said, as the last paragraph of the op says, I do wonder how much this really matters in a hot climate...
I agree - seems like there is little difference between say Ditra and vinyl. OK, the former won't get discolored - but IMO, the "avoid vinyl" warning should be replaced with "any very low perm flooring".
IMO, Joseph Lstiburek's articles (an incredible resource) should have a "submit private comments" button and he should update/clarify the article when substantial numbers of people are confused.
> put a vapor barrier under
This makes sense from a "try to keep warm side moisture out of the assembly" standpoint. Yes, it will make things worse if spilled water enters the assembly. In walls and roofs, warm side vapor barriers combined with little cold side drying are often prohibited or discouraged. Agreed, odd that this hasn't happened with hot climate exposed floors. Or moisture trapped wood in basement floors.
I know this may seem too simple to consider but... have you considered just creating an encapsulated crawlspace with insulated perimeter footings, and vapor barrier on the floor and calling it a day. Many people in the SouthEast use that method to keep their crawlspaces from rotting and would work in the other direction in keeping your floors from rotting. Add a small fan on the perimeter wall of the crawl space to pull humid unconditioned air out of the crawlspace. Have a flow through vent in the floor of the house to pull conditioned dry air into the crawlspace and then out with that fan. Is there something I've missed why that wouldn't be appropriate here? Basically it would create a conditioned dry crawlspace.
It will add ~$27,000 to the budget, not including finish work. Manual J shows that the annual savings on cooling costs would be $0, and the annual savings on heating costs would be ~$9 more than insulating the floor directly. The embodied energy costs are far greater. I also live in a very flood-prone region. There's no real benefit for me.
For anyone reading this down the road, here's some more development of the topic and my solution for my case: https://www.greenbuildingadvisor.com/question/how-to-detail-tile-to-allow-drying-to-interior?cid=238891&discussion=response#comment-238891