Image Credit: Mark Parlee Fix your leaky ducts! A comparison of the energy impacts of leaky ducts in attic space (light blue bars), sealing attic floors (orange and green bars), totally eliminating energy losses from HVAC ducts in unconditioned attics (green bars), and insulating attic floors (X-axis). In an attic with leaky ductwork, it doesn't make sense to add floor insulation above about R-19, because losses from the ducts dominate and the thicker insulation won't save much energy.
Image Credit: William Miller, Andre Desjarlais, and Marc LaFrance An unvented attic containing ductwork is a very complex hygrothermal system. The arrows depict the location and direction of both the heat and water vapor transfer of convection, conductivity/diffusion, and long-wave radiation. [From “A Hygrothermal Risk Analysis Applied to Residential Unvented Attics.”]
Image Credit: Simon Pallin, Manfred Kehrer, and William Miller This graph shows the annual variation of the moisture content in OSB roof sheathing, as shown by a computer modeling exercise. The different curves depict 224 different scenarios in a variety of climates. For most of the software iterations, the OSB moisture content varied between 13% and 17%, though several simulated scenarios result in critical moisture levels that could lead to rot. [From “A Hygrothermal Risk Analysis Applied to Residential Unvented Attics.”]
Image Credit: Simon Pallin, Manfred Kehrer, and William Miller This graph shows the OSB moisture content for the best and worst simulated scenarios for the chosen cities of U.S. climate zones 1 and 3 (the warmer climate zones). In the two cases with the lowest accumulation of moisture, the end value of the OSB moisture content is lower than the start value, which indicates that the MC must be assumed to decrease further if simulated for a longer period of time. [From “A Hygrothermal Risk Analysis Applied to Residential Unvented Attics.”]
Image Credit: Simon Pallin, Manfred Kehrer, and William Miller
UPDATED on July 8, 2015
Now that insulation contractors have been installing spray foam insulation on the underside of roof sheathing for several years, we’re beginning to accumulate anecdotes and data on successful installations and failed installations. The anecdotes and data are enough to provide a few rules of thumb for designers and builders who want to install spray foam on the underside of roof sheathing.
Increasingly, building scientists are investigating why OSB roof sheathing on many spray-foam-insulated roofs stays damp for months at a time. Most of these damp-sheathing problems involve open-cell foam rather than closed-cell foam.
I’ve been reporting on wet-sheathing problems arising from the use of open-cell spray foam since 2005, when I wrote two articles on the topic for Energy Design Update (“Vapor Retarders and Icynene,” April 2005, and “Every Failure Holds a Lesson,” July 2005). As originally understood, the problem with open-cell foam was that it is vapor-permeable, and therefore allows moisture in the interior air to diffuse through the insulation and reach the cold roof sheathing during the winter.
Five years later, Mark Parlee, an Iowa builder, wrote a seminal article on an Icynene-insulated roof with rotten roof sheathing. His article, “Repairing a Rotting Roof,” was published in the June 2010 issue of the Journal of Light Construction. One of the factors that contributed to the failure described by Parlee was high indoor humidity.
Open-cell foam under roof sheathing can be risky
At a recent building science conference in Florida (Conference on Thermal Performance of the Exterior Envelopes of Whole Buildings XII, December 1-4, 2013), two academic papers were presented that shed light on questions surrounding the moisture content of roof sheathing that has been insulated on the underside with spray polyurethane foam.
One paper discussed a field study that found that even in a relatively warm climate (South Carolina), roof sheathing can accumulate moisture when open-cell spray foam is installed on the underside of the sheathing. Researchers speculate that exterior moisture (dew or rain) between the roof shingles is forced into the OSB roof sheathing by inward solar vapor drive.
The other paper reported on a computer modeling study that showed that when spray foam is installed on the underside of roof sheathing, open-cell foam is riskier than closed-cell foam in all U.S. climate zones.
Determining the best specs for an energy-efficient attic
The first paper, “Roof and Attic Design Guidelines for New and Retrofit Construction of Homes in Hot and Cold Climates,” was authored by William Miller, Andre Desjarlais, and Marc LaFrance. William Miller presented the paper.
The three researchers ran computer simulations to determine the best specifications for attics. Among the variables considered by the researchers were:
- Adding insulation to the attic floor
- Adding insulation to the roof assembly
- Adding ventilation channels above the roof sheathing
- Adding a radiant barrier
- Specifying cool-color roofing
- Air sealing the attic floor
- Sealing duct seams
- Transforming a vented attic into a sealed attic.
The researchers acknowledged that it’s hard to come up with recommendations that apply to every situation. They wrote, “The best retrofit option for your attic depends on climate, attic geometry, duct arrangement, amount of ceiling insulation, air leakage, and thermostat setting used to comfort condition the home. If the ceiling currently has less than code insulation, savings will be greater and payback period will be shorter.”
The researchers’ modeling exercise was based on HERS BESTEST and AtticSim software. They modeled a 1,550-square-foot house with an asphalt-shingle roof with an 18° slope. The researchers assumed that the attic included supply and return ducts. They considered four different cases for duct leakage: 4%, 10%, and 20% of supply airflow.
Thick insulation on the attic floor won’t help if your attic includes leaky ducts
One of the researchers’ conclusions: When ducts are located in a vented, unconditioned attic, it makes little sense to install a deep layer of insulation on the attic floor. The duct leakage wastes so much energy that adding deep insulation doesn’t help much. (For a graphic representation of this principle, see Image #2 below.)
The authors wrote, “There are diminishing returns for adding floor insulation above about R-19 because losses from the ducts predominate. Therefore, if all one does is put more insulation down and ignore the HVAC system, then hundreds of dollars’ worth of energy is still being lost from the HVAC in the attic. Frankly, adding only insulation still leaves a poorly performing home that has higher-than-necessary utility costs month after month.”
The researchers concluded, “The most important retrofit option, whether the duct system is in the attic or not, is to seal the attic floor.” They also advise, “For new construction the best option is to keep the ducts out of the attic, make sure the attic floor is sealed to limit whole-house air leakage, and add at least the code level of insulation to the ceiling.”
It rarely makes sense to consider retrofitting a radiant barrier. The authors wrote, “Simulations for hot climates indicate that a radiant barrier can pay for itself in 20-25 years. Without leaky ducts in the attic, the payback increases to about 38-50 years. In cold climates the radiant barrier is not an effective measure.”
Damp roof sheathing in South Carolina
What about creating a sealed, conditioned attic? The authors wrote, “Sealing the roof deck and gables of an attic with spray polyurethane foam insulation has gained popularity among many builders and code officials, especially in hot and very humid climates. …. The amount of applied foam should at least match the code requirements for the attic floor. Otherwise the system may actually increase heat flows into the conditioned space, especially if there are no ducts in the attic. In fact, the procedure should not be considered unless there is a leaky duct in the attic.”
During his presentation, Miller warned against the use of open-cell spray foam to create a conditioned attic. “The roof sheathing is humid when open-cell spray foam is used,” he said.
The authors reported on a field study that raise concerns about the use of open-cell spray foam to create a sealed, conditioned attic. They wrote, “Recent data collected from a test facility in Charleston, SC (Miller et al. 2013) revealed very interesting summertime trends in attic humidity for a sealed attic as compared to the conventionally ventilated attics. … [In a sealed attic insulated with open-cell spray foam,] peaks in measured relative humidity from two different sensors in the sealed attic showed values in excess of 80%-90%, and occasionally saturated air (i.e., 100%) occurred around solar noon. In other words, the moisture content in the sealed attic was consistently 80%-100% RH from solar noon to around 6 p.m. for the seven contiguous days.”
The authors continued, “Further, the trend was observed throughout the hot summer months. … Field measurements imply that some of the moisture from a previous rainstorm migrates to the underside of the shingles and underlayment. Irradiance drives moisture from an earlier rainstorm into and through the OSB deck and open-cell foam. As a result, the partial pressure of the attic air and the partial pressure of water at the foam-to-attic air interface occasionally exceed the saturation pressure of water vapor in the attic air.”
In a recent e-mail, William Miller elaborated on the probable mechanisms involved in this phenomenon. He wrote, “We still are working to determine the mode by which moisture crosses the envelope… I believe it occurs in two different paths:
1. The sealed attic is not airtight and outdoor moisture penetrates. There is no ventilation and night-sky radiation causes the trapped humid air in the attic to diffuse into the spray foam.
2. The roof deck becomes wet during the evenings because of night-sky radiation and subsequently the condensate is forced into the roof deck by irradiance the next day.
Both modes result in moisture storage in the spray foam, and the stored water begins to damage the wood deck.”
Continuing their account of the findings from South Carolina, the researchers wrote, “On these particularly hot days, relative humidity measurements showed super saturation (i.e., partial pressure of water vapor exceeded saturated pressure). At this condition, water vapor is in equilibrium with liquid water and therefore all interior attic surfaces are wet! Without a leaky duct, the sealed attic approach actually exacerbates humidity control in attics exposed in hot, humid climates. However, the presence of leaky ducts does not necessarily provide adequate protection from attic moisture. Colon (2011) field tested open-cell spray foam in a home in the hot muggy climate of south Florida. The roof deck was protected by an impermeable underlayment and the air-handler unit and ductwork were contained in the attic, yet moisture levels in the conditioned space increased above that measured a priori sealing the attic.”
Sealed attics that include ductwork are hard to model
The second paper, “A Hygrothermal Risk Analysis Applied to Residential Unvented Attics,” was authored by Simon Pallin, Manfred Kehrer, and William Miller. Pallin presented the paper.
The researchers’ aim was to develop a computer model for sealed, conditioned attics which contain forced-air ductwork. This type of attic is complicated to model, because the temperature and humidity in such a space are affected by so many factors. (See Image #3, below.)
The authors list the following factors:
- Indoor heat and moisture production
- Hygrothermal material properties
- Natural and forced unintended air leakage
- Features of the HVAC system, i.e., dehumidifying/humidifying effect, airflow rate, etc.
- Geometrical variations of the building components
- Outdoor climate
- Orientation and location of the building and slope of the roof
- User behavior, i.e., HVAC setpoint temperatures, airing, maintenance, etc.
As the authors note, “An unvented attic hosting an HVAC system must be considered as a very complex hygrothermal system.” They decided to use two existing computer models, MATLAB and WUFI, for their study. MATLAB (a software program for mathematical computations) was used to create a numerical model, and MATLAB data were used as inputs for WUFI.
The simulated attic had OSB roof sheathing and asphalt shingles. The OSB was insulated on the underside with a layer of spray foam; both open-cell foam and closed-cell foam were modeled. The attic was modeled for seven cities chosen to represent U.S. climate zones 1 through 7: Miami, Florida; Austin, Texas; Atlanta, Georgia; Baltimore, Maryland; Chicago, Illinois; Minneapolis, Minnesota; and Fargo, North Dakota.
The R-value of the roof foam was assumed to match the minimum prescriptive requirements of the 2011 IECC; depending on the climate zone, this R-value ranged from R-30 to R-49.
In addition to the variables already mentioned, the researchers changed several other parameters to determine their effect on conditions in the attic. These parameters included:
- The thermostat setpoint
- The air leakage rate from supply and return ducts (ranging from a low of 4% leakage to a high of 20% leakage)
- The airtightness of the attic floor (ranging from a low of 0.4 cfm50/CFA to a high of 2.0 cfm50/CFA)
- Indoor moisture production.
In all, the researchers ran 224 software iterations. For each iteration, they looked at the moisture content (MC) of the OSB roof sheathing as well as the amount of energy required for space heating and cooling. (See Image #4, below.)
For each climate zone, they identified the best-performing combination of parameters — in other words, the parameters that resulted in dry OSB — and the worst-performing combination of parameters — in other words, the parameters that resulted in damp OSB.
“Open-cell foam is risky in all climate zones”
The authors wrote, “Every best-performing unvented attic roof is constructed with a closed[-cell] SPF and with a low indoor moisture supply. The opposite is true for the most risky roof. In all cases, except for climate zone 4, a high duct leakage has a positive effect on the MC of the OSB, most likely due to the dehumidifying effect of the HVAC cooling coils, which, by a higher rate of air leakage, will have a higher influence on the vapor content of the attic air during the operating cooling mode.”
As an example, the worst-performing OSB in climate zone 4 is the OSB on a north-facing roof insulated with open-cell foam, over an attic with a duct system that is fairly well sealed (4% duct leakage), an attic floor which is fairly leaky (2.0 cfm50/CFA), and a house with high indoor moisture production. After the software was run for one simulated year, the net result of these factors was OSB with a moisture content of 54%.
At his presentation in Florida, Simon Pallin summed up the researchers’ findings this way: “Open-cell foam is risky in all climate zones.”
The authors wrote, “The MC in the OSB sheathing varies mostly due to whether the spray polyurethane foam (SPF) is vapor closed or open. Having an open SPF is actually a risk in all the investigated U.S. climate zones, 1 to 7, depending on the chosen values of the other varying parameters. Naturally, the outdoor climate will influence the MC of the OSB, but also the indoor moisture production rate has a significant impact. … A high air leakage rate from the air distribution duct system has a positive impact on the MC of the OSB sheathings due to the dehumidifying effect of the HVAC unit, though it has a negative influence in terms of the HVAC energy demand.”
It’s not inward solar vapor drive
When interpreting the results of computer modeling studies, it’s important to remember that modeling results sometimes differ from field study results. That said, the work of the researchers quoted in this article consistently shows that the use of open-cell spray foam to create a sealed, conditioned attic is riskier than the use of closed-cell spray foam, in all climate zones.
Paradoxically, leaky ductwork helps protect OSB from getting too damp. If a builder chooses to install leaky ductwork in a sealed attic, the roof sheathing will probably be dryer than it would be if the ducts were well sealed.
Here at GBA, we have been receiving an increasing number of questions from homeowners who are complaining that their unvented conditioned attics (most of which were created by installing open-cell spray foam on the underside of the roof sheathing) have problems with high indoor humidity. Building scientist Joe Lstiburek has investigated a number of attics with similar problems. “How come the moisture ends up in the attic space and not in the main part of the building?” Lstiburek wrote. “Moisture-laden air is lighter and less dense than dry air. Moisture-laden air ends up in the attic due to this ‘hygric buoyancy.’” [Editor’s note: For another perspective on Lstiburek’s “hygric bouyancy” theory, see the Comment #90 by Bill Rose, below.]
Although Lstibuerk used to think that the moisture in these attics was morning dew that was being driven through asphalt shingles by inward solar vapor drive, he now reports that the inward-solar-vapor-drive theory has been disproved.
If you have an unvented conditioned attic with high humidity problems, the best solution is to install a supply-air register connected to your forced-air HVAC system in the attic (and in some cases, a return air grille as well). If the attic is heated in the winter and air conditioned during the summer, humidity problems should disappear.
Rules of thumb for builders
Here are my recommendations for builders who use spray foam to create a sealed, conditioned attic:
- Recognize that this type of roof assembly carries more risk than a vented, unconditioned attic. Keeping ducts within the conditioned space of a building (not in the attic) is preferable to installing spray foam on the underside of roof sheathing.
- If you want to lower the risk of damp OSB, choose closed-cell spray foam, not open-cell spray foam, to insulate the underside of the roof sheathing.
- To further limit your risk, consider installing ventilation channels above or directly below your roof sheathing.
- If you choose to install open-cell spray foam against the underside of roof sheathing in Climate Zone 5 or colder zones, make sure that you include an interior vapor retarder.
- If you choose to install open-cell spray foam against the underside of roof sheathing in a humid climate, your HVAC system be designed to condition the attic air and lower humidity levels in the attic.
Problems with “breathable” assemblies
Some green builders are convinced that wall and roof assemblies have to “breathe.” While “breathability” has no technical definition, many builders assume that “breathable” means “vapor-permeable.”
The argument in favor of breathable assemblies is undermined by the increasing evidence that vapor-impermeable spray foam usually performs better than breathable spray foam. A major explanation for the damp and rotting roof sheathing described in this article is the fact that the designer or builder specified “breathable” foam instead of vapor-impermeable foam.
Martin Holladay’s previous blog: “Stupid Energy-Saving Tips.”