More Musings of an Energy Nerd
In this article, I’m going to discuss a building science mystery: namely, summertime condensation near the peak of cathedral ceilings. I will propose a mechanism to explain these problems, in spite of the fact that my proposed explanation is somewhat unsatisfying.
I’m hoping that this article will prompt GBA readers to share more examples of the phenomenon — and perhaps to share more data and a better explanation for what’s going on.
Readers’ stories
As an introduction to this topic, let’s hear from six GBA readers.
Candi wrote, “Why is there condensation on my cathedral ceiling? It’s warm outside (80 degrees) and 75 degrees on the inside thermometer. The roof is in full sun all day, so it’s hot up there. There is nothing except insulation between the ceiling and roof and the condensation only forms along the beam at the peak of the roof inside. … It doesn’t happen when it’s under 70 degrees or raining. It just started when the weather got warm. We haven’t had much rain in the last three weeks, but had a lot in March and April and there wasn’t ever water [then]. The water also forms on the underside of the beams — not the actual roof side of the beams but the side facing down to the floor. The [forced air] registers are all in the floor. They run through the crawl space under the living room and between the basement ceiling and kitchen floor.”
Marty wrote, “I have cathedral ceilings on my second floor with closed-cell spray foam insulation on the bottom of my roof. There is only about a 1 to 2 inch gap between my ceiling and my hot roof. I had the roof done 4 years ago and it is completely covered with Ice…
Weekly Newsletter
Get building science and energy efficiency advice, plus special offers, in your inbox.
This article is only available to GBA Prime Members
Sign up for a free trial and get instant access to this article as well as GBA’s complete library of premium articles and construction details.
Start Free TrialAlready a member? Log in
30 Comments
One of the things I see often on TV is the remodeling show’s effect of tear down a wall and slap new tile and paint without dealing with the basic issues that all existing homes have with bad building envelopes and hvac systems, which create a rich environment for failures. I cringe every time all these “famous” tv remodelers con their clients into expensive remodels that are all lipstick on a pig, and the owner is stuck with the mess of unaddressed basic problems.
I’ve worked just on a handful of gut rehabs in my days, mainly because, in my opinion, the building envelop AND the hvac system need to be addressed first before the updates are planned. Do it right or none at all. For the record, this happens in new construction that do not address all these issues as a holistic approach.
Another summer condensation that is very common in the central and eastern South is due to indoor temperature set at 65°F, “because we like to sleep under the covers”, not realizing that their leaky house and grossly over sized hvac systems are a recipe for disasters.
Warning; somewhat long and wandering post
I don't buy the night-sky radiation theory. The water is not condensing on the underside of the roofing materials and dripping back through to the interior in most of the posted stories. From the descriptions, it is condensing directly on the interior finish materials themselves. "Marty" has a spray foamed roof with continuous ice and water shield over the sheathing. There's not much chance that any air is getting to the underside of the roofing materials. "Barbara" sees the sheen only near a skylight, and over a broad area. Water dripping back down from roof surfaces tends to happen in specific spots, along drywall seams, etc. I think the condensation is occurring at the surface where it shows up - the interior finish surface.
So the challenge is to figure out how the interior surface is dropping below the dewpoint of the air in the room. In at least some of these cases, I propose radiant cooling. With very high levels of insulation in the roof/attic, there isn't much heat getting through to the interior. So the ceiling temperature is going to be driven primarily by the air temperature inside and the mean radiant temperature of the spaces that the ceiling "sees". When there is stagnant air, radiant heat transfer can be a major component, and if there is a significant difference between the room air temperature and the floor temperature, the ceiling can be significantly cooler than the air. With floor-mounted registers and low delivery velocities, the cool air will pool at floor level, chilling the floor and furniture. The ceiling will try to match the floor temperature, even if warm air is stratified near the ceiling.
Add hygric bouyancy to the mix (I'm not all on-board with this theory, but I'll accept that it makes some sense), thermal stratification, and the possiblility that warm, humid outdoor air is getting inside and rising into the peak and it becomes easy to see how the surface temperature at the ridge could dip below the dewpoint of the local air mass.
"Rich Visconti" notices that if he aims a fan at the ridge during the afternoon, the problem goes away. Mixing and accelerating the air at the ridge would remove the warm, moist air from that space and warm the surfaces by forced convection.
This would be relatively easy to test with an accurate portable hygrometer and infrared thermometer. I actually did this two weeks ago in a commercial space. The construction was not at all like these residential cases, but the problem was similar: Condensation on the exposed ductwork hanging from the ceilings, despite the best efforts of the mechanical contractor to dehumidify and condition the space.
At floor level, the temperature was 70 degrees and the dewpoint was 60 degrees. Still a bit humid (73%) and the dewpoint was higher than the 55 degree ductwork, but past experience suggests that this close match between dewpoint and surface temperature doesn't produce the volume of condensation that we saw in the space. We went up on a lift and felt a significant stratification layer just at the level of the ductwork, about 10' below the ceilings. Up above the ductwork, the temperature was 82 degrees and the dewpoint was 71 degrees. This is a whopping 16 degree difference between surface temperature and dewpoint. In this case, it turned out that the building is running at a negative pressure and there are tons of outdoor air leaks. The outdoor air was 95 degrees and had a 74 degree dewpoint (yecch), and this air was coming inside, rising to the ceiling and staying there. I realize that this is a forced cooling example rather than a radiant cooling example, but it shows how different the conditions can be between the floor and the ceiling in a tall space, with substantial differences in absolute humidity as well.
It would be easy to test this theory in one of the houses that experiences these issues. Measuring air temperatures, dewpoints and surface temperatures would immediately show if the interior finish surface is cool enough to condense the local air mass.
I'm with you on this, Martin. The surface where condensation is occurring doesn't have to be "cold"...it just has to be colder than the dewpoint of the indoor air. There's typically at least a little bit of thermal bridging at a ridge beam that would yield slightly colder surface temperature at the inside peak overnight and early morning, especially on a clear night. Combine that with a house with high dewpoint air, and the tendency for the indoor dewpoint to be higher at the peak, and you could get an occasional case of condensation.
Regarding solutions - opening up the ridge seems like the hardest option to me. I think the most straightforward solution would be to reduce the indoor dewpoint by using a whole-house dehumidifier, optimize the existing AC system (if present) for better moisture removal, AND use a ceiling fan (or other fan) to destratify the air.
I admit I don't know enough about cathedral ceilings or condensation to dispute how much thermal bridging is actually there, and how small a temperature change would be required for it to condense. But surely, if you propose the mechanism is thermal bridging, then air sealing is not only the hardest option, but entirely pointless.
I didn't mean to imply air-sealing as a solution...the "open up the ridge" solution in my mind would entail adding enough exterior insulation to warm up the indoor surface temperature above the dewpoint. As you can imagine, this wouldn't be easy...hence my comment.
Peter's explanation sounds more plausible. Humid air escaping into the attic with a cold roof would condense inside the attic, and I can't see how it would contribute to condensation inside the conditioned envelope.
For condensation on the interior surfaces to be caused by night sky cooling I think would require a substantial thermal bridge between the roof exterior and the condensing surface. I didn't see anything suggesting that was the case in the examples given. And if it was the case, sealing air leaks wouldn't help.
Its probably unlikely, but I wonder if the people experiencing this problem all have light colored roofs? On a hot, humid day its conceivable that a white roof for instance could radiate enough to heat to cool the ridge board/beam down enough to condense.
Rick,
Can you clarify what you mean by "radiate enough to heat to cool"?. It almost seems like you're suggesting that the exterior side radiating is going to pull heat from the interior side. I'm pretty sure that's not possible.
Rick,
It's more than unlikely. The daytime sky temperatures are too high pretty much everywhere to get any significant radiant cooling of the roof in daytime. Plus, you have the solar heating of the roof that will more than offset any radiant cooling for net roof temperatures much higher than ambient air temperatures. "Cool" roofs are only called that because every other roof is so hot.
Night-time radiant cooling of the roof plus thermal bridging might bring the indoor surface temperature of a ridge beam down below the rising dewpoint of daytime air in some of the listed cases, but I'm not convinced that this is the mechanism for most of them.
Condensation transfers quite a bit of heat, warming the condensing surface. To get enough condensation to actually drip, I think you've got to have some sort of active cooling - not just leftover cool temperatures from the night before. Cold air chilling the floor and the floor actively cooling the ceiling by radiation could do it.
Peter, Trevor;
I believe I read a report from Building Science Corp about commercial buildings in the desert southwest that were experiencing roof rot after simply changing their roofing membranes from black to white. Despite the high temperatures and dry-climate, these roofs were rotting from condensation.
After researching it, they determined that the white roofing material was reflecting so much heat that it was acting as a kind of cooling coil and warm (interior/humid) air was condensing on the under-side of it. I believe they stated that had they simply switched to a black roof, the absorbed heat would have been sufficient to prevent the issue.
I wish I could find the link but my computer isn't letting me access the building science corp website for some reason... I'll try again later.
The article was from GBA...
Here is the link: https://www.greenbuildingadvisor.com/article/insulating-low-slope-residential-roofs
From the article:
"Uncertain of the cause of the drywall cracks, one of the builders called in William Rose, the well-known building scientist from the University of Illinois, to investigate. Rose discovered that the homes had wet roof sheathing — due in part to the type of roofing installed on the affected homes (white membrane roofing). “In December, January, and February, the fiberglass was wringing wet,” Rose told me. “In this climate, radiant effects become really important. There is nothing standing in the way of the roof radiating out to space. You have a whole lot of heat loss from the roof surface, day and night. With this white roofing, 80 percent of the heat that hits the roof is reflected. The sun can’t keep up with the heat losses to the sky. What you’ve created is a sky-powered cooling coil, and the fiberglass insulation is like a dirty condensate pan. The roof sheathing gets so cold that it is sucking wetness out of dry air.”
Obviously, there is a lot more going on here- including putting air permeable insulation in the attic. This wasn't the case with some of the people having issues as described in the article. But given that ridge beams or ridge boards almost always constitute a thermal bridge to the interior (unless there is overlapping exterior rigid foam) perhaps it is possible that spray foam wasn't enough?
Again... its a long shot, I admit.
Rick,
I think the article is a little bit misleading about the'day and night". I agree with the author completely, in that 80% of the heat that hits the roof is reflected back. That's the point of reflective roofing. But it's not getting colder than the air and condensing moisture during the daytime. That only happens at night when the sky temperatures are really cold. But with the reflective roof, the daytime heating is not enough to drive out the moisture that accumulated during the night. As a result, the moisture just kept building up in the cavities. Even in Arizona, the daytime sky temperatures aren't low enough to cool the roof below the air temperature.
Peter,
The temperature of the air and the temperature of water vapor in the air may be warm in Arizona during the day, but the temperature of outer space is very, very cold -- 24 hours a day. In cloudy weather, the warm clouds restrict radiation of heat from the earth to outer space. But in clear, dry weather, objects on the earth radiate heat to cold outer space -- even during the day.
Countering the effects of this radiational cooling is sunlight, which warms objects by radiation during the day. But very reflective white roofing reduces the heating effect of sunlight. Here's how building scientist Bill Rose explains the phenomenon: "‘With a clear sky, there’s radiation loss to the sky, and it happens 24 hours a day. A clear sky is always really cold. ... At the house we’re monitoring, it appears that perhaps solar heating can’t even compensate for the daytime losses.”
More info here: "Night Sky Radiation."
I certainly agree that roofs are always losing heat to the sky, and that space is cold. i also admit that I have not spent much time on white roofs in Arizona, so my direct experience is limited. But I have spent quite a few hours on white roofs in the northeast, and I can tell you that they are still hot, or at least warm, even on a cold, clear day. This is because, at least around here, there is always enough moisture in the air to reflect some part of the earth's escaping heat back at the ground. I was somewhat careful in my above posts to hedge the language a bit, saying that in most areas, solar heat gain is going to exceed radiative heat loss during the day. Bill Rose's data from Arizona may show that this is not always the case, and that is certainly interesting and instructive.
But it doesn't change my overall thoughts on the indoor condensation issues. You did not mention the locations of the houses in your article, but I suspect that they are not all in Arizona or other high altitude desert climates. And even in the houses that Bill investigated, he wasn't finding condensation on the interiors in the afternoon - he found condensation within the ceiling cavities, apparently on the underside of the chilled roof where we would expect it to be. Not at all the same animal.
I'm keeping an open mind, but I'm still not convinced. I would love to measure the conditions in one of these houses when the condensation is happening. If I run into one locally, I'll let everyone know how it turns out.
The one thing I see in common is unvented cathedral ceilings with recent changes in airflow or vapor diffusion.
We need to help people understand that unvented cathedral ceilings are risky assemblies. If you have one with a history of working think long and hard before you change anything!
The changes all sounded logical and reasonable but seemed to have disturbed the delicate balance of their risky assemblies. I do not see quick, easy or cheap way to fix the newly created problems. I do not think any of us would recommend building the current assemblies. What can we recommend put it back the way it was? Know one can guarantee that Humpty Dumpty will work the same after you him back together again. Tear off your roof and rebuild it with better air sealing, more insulation and ventilation? That too expensive so that will not happen.
If night sky radiation where the main force behind the phenomenon, it seems to me the complaint would be that the condensation accrued between 10 PM and 10 AM.
Walta
I haven't had time to study all of the information here but I did experience summer condensation problems in a geodesic dome (which is sort of a cathedral roof). This assembly was definitely risky but held up just fine for 43 years and counting. The outer shell was perfectly sealed (fiberglass) and there was a vapour barrier on the inside surface with 3" foam as insulation and a 1/2 inch air gap at the outer surface. In summer, the air in the assembly basically super-heated and under some conditions would actually blow steam out of a poor seal at the peak. The surrounding area was wet. The problem was "solved" by carefully sealing the peak.
My feeling was that the assembly slowly acquired moisture through the gap when it was cool and then released it when it was hotter.
Keep those anecdotes and theories coming! It's all good.
I appreciate both Peter Engle's point (his observation that the descriptions seem to refer to moisture that forms on interior surfaces rather than water that is dripping through the roof assembly) and Rick Evans's point (that white roofing might be part of the puzzle).
One of the main purposes of this discussion is coming up with a series of questions to refine our understanding of what's going on when this happens.
I have no personal experience with this phenomenon but my first thought is that it is somewhat more common than we think and goes unnoticed by most people. Roof color may have something to do with it but my feeling is that a key factor, and probably the simplest solution, is air movement. Inside corners have trouble with air movement when they're vertical, a cathedral is a kind of worst case inside corner. Stratification is the only force acting on the pocket of air up there and it drives heat and moisture into it until the dew point is above surface temperature. This may be the one reason to run a ceiling fan when nobody is in the room.
Good post , starting wth readers' concerns. and good comments. So far, it sounds right to me. Energy efficiency measures lead to short-cycling of AC and higher indoor humidity than in the past. It seems to happen on clear nights, so it's pretty clearly sky radiation cooling the ridge of the roof. The detailing right at the ridge can get messy with voids and SIPS panels and doubled ridge beams, so it's hard to tell exactly which surface is condensing.
I'd just like to add one thing. Martin, you touched on it in a previous post, Feb 3, 2017, about the Building science To-Do list I presented at the Clearwater conference. You quoted me as saying “Corners are some of the most interesting parts of the building,” said Rose. “I did a survey of freeze-thaw damage on brick buildings. Not much damage shows up in the field of the brick except where the brick is wet. Corners get wet because they stick out; they are exposed. And modifying heat loss at corners is difficult to do. The radiant environment is really strong at corners. Pointy things can drink from the air.”
Ridges have essentially zero air film resistance because there's no stable cushion of air, and because the radiant exchange with the cold sky is particularly strong. Here's an image I use to illustrate the cosine rule. Basically, the exchange between a surface and the sky is a function of the difference in absolute temperature to the fourth, modified by the surface emissivity and the cosine of the angle from normal to the plane. The ridge sees a whole lot more sky than the flat of the roof, and the sky can get cold--you can shoot it with a pyrometer or an IR camera.
In theory, the ridge is cold. Is it really cold? I don't know, we need to take IR shots from inside and from above, and it hasn't been done systematically as far as I know.
A note about "hydric buoyancy". I use the term with caution. At equilibrium, gases are very well mixed, and for the height of a building, the difference in humidity top to bottom is detectable only at the fourth decimal place. I like to point out that if gases really did stratify by molecular weight then I'm a billionaire because all I need for commercial gas separation is patience. But as a transient condition, yes. Ask any cumulus cloud. Ask how a fogged hotel room mirror clears up. And the pingpong effect that BSC talks about is quite correct I think.
Bill,
Thanks very much for weighing in.
I'm well aware of your reluctance to use the term "hygric buoyancy," Bill. In the context of this article, it's a shorthand reference to the fact that indoor air near the peak of a sloped roof is often more humid than air near the floor.
If any GBA readers want to know more about the "hygric buoyancy" controversy, they should read "High Humidity in Unvented Conditioned Attics."
I don't think I saw mentioned whether there was a consistent ridge detail between projects. In cathedral ceilings, there is often a large thermal bridge where a structural ridge spans from the exterior down into the interior. Same around skylights; people often overdo the framing around them, when single headers are almost always strong enough, but I don't recall ever seeing single headers spec'd at skylights. Combined with night sky radiation and the other factors that Bill Rose listed, it seems like the ridge beam is funneling heat away from the house. (There is an easy solution for new construction--I try to drop heavy ridge beams below the rafters, so insulation can continue over the top.)
In addition to this, assuming a hand cut roof, there may be blocking between rafters thereby increasing the amount of wood (ie, making the thermal bridge wider).
I realize this is an older post, but I ran into this cathedral ceiling condensation issue today. Outdoor air temperature when I noticed the condensation was around 50°F with a relative humidity of around 30%. Indoor temperature was in the mid 60's. Bright sunshine all day and the home has a dark colored roof. I saw the condensation around 3 PM. The home I am working on is log, which has been for the most part gutted. We are just starting to re-frame interior walls, using LSL framing. No one currently staying in the home. Lots of traffic in and out of the home while working on it. Heating system was turned off around 10 am, not warm enough for AC use. The area where the moisture was found is around 14 feet above the floor on the west facing roof. There has been no rain for a few days, snow has been gone for several weeks with nighttime lows around 30°F. Unfortunately I did not have any of my energy auditing gear, would have been interesting to see the area with a thermal image camera. I will have it with tomorrow, just in case. I also do not know what the interior humidity level was. First time experiencing this phenomenon, interesting and puzzling.
Randy,
Well, it's certainly getting cold enough at night to cool the ridge -- especially if nights are clear, and night sky radiation is a factor. If you have a different theory, please share it.
I will be taking my test equipment with today, I'll take some temperature and humidity measurements. Maybe we'll be lucky and it will present again. I don't have a good theory, moisture was on west facing roof in full sunlight at 3 PM in the afternoon. HVAC thermostat indicated house was in the mid 60's, outdoor temp of 50. Roof has a low R-value, (1 1/2 inches of Polyiso and 3 1/2 inches of rockwool batts), not well air sealed. There is a ridge and eave vents, but no clear path, air has to move through the insulation batts. I did touch the roof to see how much moisture was present above the visible water drops, roof did not feel hot or cold, but my hand was wet after. The property is lake shore, could have some effect.
The condensation at the peak of the cathedral ceiling of the log cabin I am currently working on returned yesterday, this time I had my test gear with. The humidity level at the ceiling was 87% with a temp of 73°F. Humidity near the floor was 35% with a temp of 68°F. If I calculated correctly, the dew point temp of the ceiling was 71°F. Thermal imaging indicates the temp of the ceiling was right around that temperature. What is still puzzling me is why the elevated humidity at the peak? My hypothesis is it has to do with the roof assembly. From the outside in, shingles, #15 felt paper, plywood, 3 1/2" rockwool batts (no air space), 1 1/2" polyiso foil faced, another layer of # felt, 1 1/2" tongue and groove paneling, not well air sealed. I stated in a earlier post, the roof has eave venting, but I have discovered there is no ridge vent. Air and humidity enters the eave venting during the evenings and slowly migrates to the ridge where it becomes trapped due to the low perm roofing and no ridge vent. As the roof warms during the day, the accumulated moisture finds a path to the inside where it condenses on the tongue and groove paneling at the peak. The condensation is only present on the west facing roof in the areas that have the eave venting, no condensation in the areas where there are no eave vents due to the roof design (see picture of home.) Moisture on the ceiling seems to be present only after noon. I think the addition of a ridge vent may eliminate the issue, but will also increase winter time stack effect. Home is located in climate zone 7, approximately 10,000 HDD.
Randy, there is a possibility that Winter condensation remains in the rockwool bats and it takes several weeks to evaporate. In the spring, the west facing roof rockwool bats warm up enough to transport moisture up and inside as there is no ridge vent. Air indoor may still be cooler than outside and still be sinking, pulling air from the eve vents through the moist heated roof channel.
I have a similar cathedral ceiling situation (eve vents and no ridge vent) where on hot and humid days, with AC on, the cool AC air sinks and pulls air in from ceiling ridge and eve vents, assisted with roof heat and air rising in the channel between rafters. Not sure which contributes more: sinking AC air or rising warm roof air, but the result is transport of humid outside air that condenses on cool ceiling surface inside. I get similar thermal camera images as you do, including telltale air infiltration pattern and ceiling condensation droplets near the ridge beam air gaps. Some ceiling mold also appears after a while up to a foot away from those spots.
I upgraded attic insulation over the winter, resulting in less a/c use than I'd typically have this spring. Indoor humidity was high. I had condensation around a ceiling fan at the peak of a cathedral ceiling, the highest point in the house, the only ceiling penetration in the room. It was happening at midday, when the roof would have been well over 90 degrees in full sun, so the overnight cooling explanation probably does not apply to my situation.
I believe the cause was humid indoor air being pulled into the walls at multiple wall penetrations and rising through loose fiberglass batt insulation due to the chimney effect. The upper walls and roof are sun-exposed, so there would have been air-heating and expansion. With only one penetration in the cathedral ceiling, that warm moist air intruded into the room, encountering cool drywall ceiling and metal fan housing, resulting in condensation.
I air-sealed the outlets and switchplates and everywhere else I could think of, especially including the area around the fan -- eliminating the exit point and as many entrance points as I could. I also reversed the ceiling fan to get some circulation up there. The problem went away. I'll need another spring to see if it happens again.
Two years ago I finished a project which included turning my attic into living space, adding solar on top of a vented copper roof and ridding myself of heating oil by adding heat pumps. The attic is heavily insulated with foam and roxul in various combinations. The ceiling (specifically in the ridge area) is dropped about 18” below the original ridge and packed full of roxul.
This afternoon I walked upstairs and noticed condensation on the top 10” of the shower door in the attic bathroom. Across from the shower the mirror had condensation on the top 3” or so. Looking up I saw the underside of the painted wooden ceiling was wet. In the bedroom area, the same thing had occurred.
I do own a portable hygro thermometer. The AC was on and the room temp was 71 degrees (my kids must have been messing with the tstat) and the humidity was 63%. It was 90 degrees outside and the time was 3pm here about an hour north of NYC.
Though I felt there was an issue with the heat pumps last summer as the upstairs had felt cold and clammy and humidity levels hovered in the 50s, I hadn’t really felt that this summer.
I am perplexed. Perhaps a whole house dehumidifier is in order. Or I need to balance the air handler better (though in the attic, the return duct is almost at ceiling level and I have one supply across the room at same height)..
The night sky radiation theory simply can’t be possible in this situation. The glass mirror and shower door are feet away from the roof. I imagine the heat pump is oversized (though an engineer did manual j calcs) even when it’s variable speed is at its lowest.
Perhaps I just aim a fan at the ceiling and hope for the best? Turn off the AC?
Ted,
A fan might help. It's hard to determine what's going on without a site visit.
We need to start with the basics: the type of condensation you describe occurs when humid air contacts a cold surface. So that implies that your surfaces are too cold, or your air is too humid.
An indoor relative humidity of 63% is high, so one possible solution is to install a portable dehumidifier in the room with the problem.
Another possible solution is just making sure that everyone in your family runs the bathroom exhaust fan when they use the bathroom.
A third possible solution might involve determining why the top of your shower door and the upper part of your bathroom mirror are the coldest surfaces in your bathroom. One explanation might be that the diffuser or register in your bathroom aims the conditioned air from your air conditioner at these surfaces -- so perhaps a different diffuser or register might solve the problem.
Log in or become a member to post a comment.
Sign up Log in