This question is borderline off topic, but since this place is such a great resource of brilliant minds, I was hoping you all could help me out. We're having issues where I work with condensation forming somewhere in the insulated space between the ceiling and roof, primarily on hot, sunny days. My boss is working on some plans for ventilating things, removing humidity, etc. and we've also put sensors in the ceiling area and in the room in order to monitor conditions and see if we can find a clear pattern as to when, why, and how this is happening. The sensor setup is pretty straightforward - we've got two AM2302 (wired DHT22) temperature-humidity sensors connected to an Adafruit Feather M0. The wiring is literally just 3.3V supply, ground, and one signal wire from the the Adafruit Feather up to each sensor. According to the sensor datasheet, no external circuitry is required. Readings from the sensor in a normal indoor room environment are predictable and consistent with what we'd expect.

My question for you guys is about some readings I'm getting that are strange to me. The ceiling sensor is placed almost directly against the plywood panels that make the structure of the roof. There is fluffy insulation (fiberglass I think, but I'm not sure) below the sensor for the full height of the ceiling joists (maybe 10-12 inches,) then below that is a thin (maybe 1/2 - 3/4") rigid foam piece of insulation with foil sides which forms the bottom of the insulated area and the ceiling of our shop (the rigid foam separates the mostly trapped, insulated air space near the roof from the ambient air in the room.)

On sunny days, the temperature at this sensor, which is essentially touching the roof, climbs rapidly as one would expect. The relative humidity reading also drops rapidly as one would expect. However, I'm also calculating the dew point from the temperature and relative humidity readings, and the dew point doesn't stay stable like I would've expected. Instead, when the roof is heating, the dew point is climbing rapidly as well. Given that there aren't significant air pressure changes happening during this time, the dew point should be a pretty close approximation of specific humidity and should not change at all as a function of temperature. It should be quite stable unless the actual moisture content of the air is climbing. Dew point calculation is from http://www.ajdesigner.com/phphumidity/dewpoint_equation_dewpoint_temperature.php#ajscroll:

And here's the data from our sensor over a weekend:

My boss' theory is that heat from the sun is driving moisture out of the plywood roofing and into the void, driving up the specific humidity. I wouldn't expect the wood to be holding enough moisture that it could have this effect, but I don't have any better explanation. My only other thought is that the sensor is misbehaving, reading either the temperature or the relative humidity in the space significantly wrong, throwing off the dew point calculation.

Do you guys have any experience with temperature humidity sensors like this? Does it seem likely that it would give really bad results in a partially enclosed ceiling space? Does the theory of wood panels releasing moisture and driving up the specific humidity/dew point make sense?

Just to be clear, I know there are lots of possible causes and effects for our condensation issues, and lots of ideas for solutions that may be better than what we're working on. For the moment, I'm not worried about any of that. All I want to know is whether or not I can trust the data I'm getting, and where that moisture is coming from if there really is a legitimate spike in moisture when the roof heats up.

 

My boss' theory is that heat from the sun is driving moisture out of the plywood roofing and into the void, driving up the specific humidity. I wouldn't expect the wood to be holding enough moisture that it could have this effect, but I don't have any better explanation.

That's what I was thinking, prior to reading you boss' theory. Plywood can hold a lot of moisture especially if the roof is leaking.

 

The most common failure is the basic structure. Things like, gable end vents or "twirleys" (on top of the roof). Proper removal of local humidity is created by laminar flow in every rafter space. This is accomplished with soffit vents in every rafter space and a 100% length of ridge vent on top of the roof. Gable vents and twirleys distort the flow so it is NOT laminar.

(Geeze, didn't anybody tell you to check with Number Twelve for environmental air control?)

 

The most common failure is the basic structure. Things like, gable end vents or "twirleys" (on top of the roof). Proper removal of local humidity is created by laminar flow in every rafter space. This is accomplished with soffit vents in every rafter space and a 100% length of ridge vent on top of the roof. Gable vents and twirleys distort the flow so it is NOT laminar.

(Geeze, didn't anybody tell you to check with Number Twelve for environmental air control?)

Somehow I had a feeling you'd have some answers! Thanks for your insights.

Ok, so I can picture that pretty clearly for a typical residential roof, but what about a nearly flat roof? It's got just a tiny, tiny bit of pitch to it, but it's basically flat, and the rafters are perpendicular to what little slope is there. Sadly, they're also parallel to the only side of the building that has any overhang with soffit area that could be vented, so the best we could do there is vent those couple rafter gaps, but not the remaining 95% of the building. We've got no ridge, nor gables to vent. Finally, the space between the rafters is filled, top to bottom, with insulation. So there is essentially zero airflow potential. My boss has elaborate plans for putting a giant duct down the middle of the shop, intended to draw air (through the dense insulation) from all these rafter spaces into the central duct and then dump that hot, moist air outside. Among my top fears with this plan is my belief that if one rafter space is less thoroughly plugged up with insulation, air will take the path of least resistance and flow primarily through that one, leaving all of the other spaces still hot and damp.

Do you have any thoughts on the sudden spike in measured/calculated dew point when the sun beats down on the roof? Do you agree with strantor that the plywood could easily be retaining lots of moisture and then releasing it when heated? I was originally skeptical about this idea, but the data so far appears to back it up (and of course, I'm inclined to trust strantor... but a second opinion still wouldn't hurt!) When I first saw the data I thought I was having sensor issues, because I didn't think there was any explanation for such dramatic swings in dew point, but now I'm not so sure.

 

Do you have any thoughts on the sudden spike in measured/calculated dew point when the sun beats down on the roof?

I'm with strantor. That's why I didn't say anything about this post except to click a "like" on it.

You have successfully described the real problem. You can buy corrugated (wiggley) Styrofoam spacers to keep the warm fuzzies from blocking the air flow from the soffit and create a laminar flow just barely below the rafters.

https://www.amazon.com/Attic-Baffles/s?ie=UTF8&page=1&rh=i:aps,k:Attic Baffles

That still leaves the rafters rotting from moisture cycling. How you cause a chimney effect on the, "high" side is about new structure. You need to connect all the vent spaces to a plenum (in a very broad sense of the word) and add a vertical pipe (or pipes) to create a chimney effect which will pump the air flow for free. The open vent area should be at least 1% of the total roof area. The square footies of the vertical pipe should be at least 10% of the intake area of free flow in the spacers. (For instance, if the baffles are 1" spacers, 100 linear feet of roof needs pipe I.D. of 3.2 inches and follow the 10x2 rule. The chimney should be 2 feet higher than anything within 10 feet of it laterally. More "up" means more air flow because chimney effect is driven (linearly) by delta temperature. You don't need a hurricane in there, you just need persistent, reliable flow.

A "real" roofer might know a better way.
Your structure is wrong in the first place.

How about entrance holes in one end of the flat roof, and the rafter spaces provide channels, if they are not blocked in any way? Then do a plenum and chimney on the other end?

Got the picture?

 

Your structure is wrong in the first place.

Amen to that!!!

I must confess that I hadn't given any thought to the consequences of our goofy roof structure until your phrase "laminar flow in every rafter space." Then it clicked and I realized we were pretty well screwed!

Sigh... I wish our boss had talked to an expert when they bought the property. It didn't have any of this insulation when they bought it, and he put it all in himself, thinking he knew what was needed. With good professional advice, this could've been done right the first time, probably with half the labor and half the expense of working on it twice!

And, sorry for the redundancy asking for comments on strantor's opinion. I didn't see your "like" there.

Thank you both for your insights and advice!

P.S. I feel like I should be taking notes on all this for when my wife and I buy a house sometime in the next few years. Some handyman stuff I'm quite good at, but roofs, insulation, ventilation, etc. are all new to me.

 

Thank you both for your insights and advice!

You got schooled.

 

Regarding the apparent rise in dew point, I’d be happy to check your calculations if you want to give me some combinations of your sensor readings, say from the peaks and valleys of your readings. I can see if I get the same dew points you do. I’m developing an app and one feature it has is a complete psychometric calculator.

 

give me some combinations of your sensor readings,

That data is in post #1
give or take how well the graph can be read.

 

That data is in post #1
give or take how well the graph can be read.

You underestimate my laziness!

 

Regarding the apparent rise in dew point, I’d be happy to check your calculations if you want to give me some combinations of your sensor readings, say from the peaks and valleys of your readings. I can see if I get the same dew points you do. I’m developing an app and one feature it has is a complete psychometric calculator.

Thanks for the generous offer! Here are a few data points towards two extremes:

Temp(F)....RH %....Dew Point(F)
..128.12..........26.....................83
..129.74..........26.....................85
....67.46..........40.....................42
....56.12..........61.....................43

 

Thanks for the generous offer! Here are a few data points towards two extremes:

Temp(F)....RH %....Dew Point(F)
..128.12..........26.....................83
..129.74..........26.....................85
....67.46..........40.....................42
....56.12..........61.....................43

I get 82.5, 83.8, 42.3, 42.9. So your calculations appear to be working and are not the source of the anomaly.

 

I get 82.5, 83.8, 42.3, 42.9. So your calculations appear to be working and are not the source of the anomaly.

Awesome! Thanks for the verification.