I've gotten many calls over the years about attics and attic ventilation; almost invariably the caller is confused, having heard different things from different people. In this article, I'll discuss the performance of attic assemblies and try to shed light on why there are so many points of view about roof ventilation.
The temperature of a northern-climate roof we monitored throughout the 1990s is shown below. Here is a summary of the study: The roof gets cold at night and is hot during the day. It gets hotter on a sunny day than on a cloudy day. Attic assemblies with openings to the outdoors ("vented" attics) stay a bit cooler during the daytime than unvented assemblies. They also stay slightly warmer at night.
Sheathing temperatures are affected somewhat by roof ventilation, but many other factors play a bigger role.
Many factors influence the temperature on the roof. A prioritized list might include hour of day, outdoor air temperature, cloud cover, color of the roof, roof orientation, where the measurement is taken (sheathing or shingles, top or bottom), latitude, wind speed, rain or snow on the roof, heat conduction across attic insulation, roof framing type (truss or cathedral), and attic ventilation to the outdoors. As you can see, ventilation falls pretty far down the list.
To better understand how wind affects roof ventilation, Canadian researchers T.W. Forest and I.S. Walker measured the air exchange rate in attic assemblies using tracer gases. The graph below (Figure 2) gives us a feel for what they found. That is, air-change rates in the attic tended to increase with wind speed, but the amount of air change at a given wind speed was unpredictable. In fact, even with specific information about climate, construction type, and wind speed and direction, the resulting air-change rates may vary by a factor of 10 or more. Whether air flows out through a roof opening or in through that opening, and whether this airflow induces flow from indoors into the attic or helps dilute and remove moist air from the attic, can never be pinned down very well, except to say that wind is a more powerful factor than buoyancy (the "stack effect").
While higher wind speeds tend to increase attic ventilation, the relationship is a weak one: Ventilation rates at a given wind speed can vary by a factor of 10.
For the most part, roof assemblies behave like any wood structure — they are wetter when cold and drier when warm. Roof assemblies tend to be hot, thanks to the sun, so they tend to be dry. Of course, if the roof leaks, that becomes the biggest source of wetness. High moisture levels indoors or in basements or crawlspaces can also increase moisture levels in the roof. Roof members can become particularly wet or covered with frost near holes in the ceiling or leaks in attic ductwork, where humid air enters the attic. It was the formation of local frost "walnuts" like those shown on the next page that led researchers in the late 1930s to recommend attic ventilation. (If only they had offered to seal up the ceiling instead!)
Many attic assemblies are built with vents to the outdoors on the presumption that outdoor air will enter the attic and dilute moisture coming from indoors or from the foundation. The further presumption is that indoor air is wet and outdoor air is dry. Both of these assumptions are often false. If there are openings in the ceiling, then air movement in the attic can induce airflow from below, or dilute air from below, or do nothing, in ways that are just plain unpredictable no matter how much research is done. Attic air movement can also induce flow into the living space below, which is a nasty problem when the air conditioning is running.
Moist interior air leaking through a hole in the ceiling can produce moldy sheathing or frost on a roof truss. This photo by the author shows results from the Attic Performance Project
Observations in the Field
Suppose that the picture of attic ventilation provided by physics, described above, doesn't quite cut it. Too many qualifications; nothing pinned down. Then we can go to our own observations and experiences, subjective and incomplete as they may be. Here's my main finding: Attic assemblies built over the last 15 years or so are pretty good. They may be a crapshoot in building-physics terms, but the crapshoot is heavily biased toward good performance.
Let's look at attic assemblies by component:
Truss construction seems to do quite well. There are disasters that occur during construction. Truss uplift continues to be a problem requiring cosmetic fixes. The industry has, for the most part, discontinued the use of fire-retardant treatment of truss members, thereby avoiding what was a serious concern for several years. The truss heels in many cases still fail to provide the height necessary for good insulation. Attics have become a forest of truss webs, and thus are less usable for attic storage space. But the overall picture is good (at least by my observations).
Gypsum wallboard ceilings have shown improvement. The message seems to have gotten out that ceilings must be airtight — there is no justification, summer or winter, for allowing indoor air or foundation air to pass into attic cavities. The common culprits, such as framed soffits over kitchen cabinets, open oversized plumbing or mechanical chases, and leaky can lights, are going away in most construction where the word has gotten out. Weatherization of existing buildings has kept the focus on closing off any ceiling bypasses. In my experience, most truss-framed attics do fine without special vapor-barrier membranes in the ceiling, but in cold locations, cathedral ceilings may need vapor protection just as walls do.
Insulation. Regarding insulation, most areas of the country have healthy amounts in the attic — R-30 in general and R-38 in northern areas. Cellulose provides good insulation and helps block airflow. Fiberglass, in sufficient density and with good installation, also provides good thermal insulation. Foam insulation is being used more commonly, and has become the material of choice for residential air-sealing. Structural insulated panels (SIPs) work fine, as long as the airflow problem at joints is addressed. Foam insulation has been sprayed on the underside of board and wood-panel sheathing with great success. Insulated panels (often polyisocyanurate) make for good roof-deck assemblies, as we know from commercial low-slope construction, where the foam insulation is often sandwiched between the structural roof deck and the roofing membrane. (All foam needs fire protection, of course.) Open-cell foams such as Icynene may need more vapor protection than closed-cell foams, which have greater resistance to vapor flow.
Vapor barriers still cause squabbling, but most builders know that moisture flow from below comes mostly through holes in the ceiling. Cathedral ceilings require special care in insulation placement and vapor protection. But the new code provisions should encourage insulated sheathing materials or insulated "sandwich" assemblies that resist moisture transport and heat flow as a package. With these roof assemblies (I call them "insulated vapor retarders" or "fat vapor retarders"), the inside surface stays close to indoor conditions, the outside surface stays close to outdoor conditions, and nothing bad happens in the middle. Our laboratory has had such an assembly in place for more than 15 years, with one inch of foil-faced polyisocyanurate insulation directly beneath the OSB decking; the sheathing gets hot during the day, but the OSB above the foam insulation is the driest sheathing of all. Remember: Hot means dry.
Ductwork in unconditioned attic assemblies is not ideal. It's best to place all ductwork in conditioned spaces.
OSB has become the universal sheathing material, by economic and environmental necessity. But we still know too little about the moisture performance of this material, such as under what conditions it will begin to fail. In my laboratory, we have seen the material swell by 50 percent or more under extreme conditions. Will it begin to show signs of sagging between trusses, or will workers be putting their feet through it at the time of reroofing? I don't know, but the absence of signs of product failure in the field — at least to my drive-by observations — is reassuring. Nevertheless, I look forward to the day when the marketplace provides a product with more clearly established performance characteristics. I'll be a strong supporter.
Shingles. I'm reviewing the condition of the shingles installed on our research laboratory in 1989; after 18 years, signs of aging are appearing. We hope to conduct laboratory tests to pin down and better quantify the shingle performance and the factors that influence it. The aging we see shows some temperature effect: The white shingles are in better shape than the dark, and a few of the most aged-looking shingles are found on the hottest bay, the one with foam directly on the underside of the sheathing. Without the numbers to go by, we must rely on observation, and our observations suggest that performance depends on other factors besides the presence or absence of ventilation and whether the assembly is truss-framed or cathedral ceiling.
Of course, natural weathering tests that began 18 years ago say little about shingles that are made today. I sense that the shingle industry is currently producing dimension shingles that seem to lie quite flat, resist wind uplift, and hang on to their UV-protecting granules. I don't know how to reroof over dimension shingles, and it does seem unfriendly to the landfill to have that much more mass in the shingle. Nevertheless, my drive-by observations show a lot of good-performing shingles going on roofs over the last couple of decades, and that is very reassuring.
Roof vents. Many years ago, we measured the "net free area" of about a dozen ridge-vent materials. (We used an apparatus that measures the pressure drop across a vent device with great accuracy.) We found that ridge vents with large openings (minimum opening dimension around 1/4 inch) had an equivalent net free area very close to their rated capacity. Vent devices with small openings — or with filter fabrics, or scrims — performed much worse, as much as 75 percent less than their rated area. (If you want to know how restrictive a vent device is, use your imagination — if it looks like air would have a hard time moving through, it probably does.)
This discrepancy would be a big deal, I suppose, for someone who felt that vent regulations were critical to attic performance. I don't, so for me, having vent devices with less airflow than advertised is not a cause for concern.