Architectural Record BE - Building Enclosure

Roofs and Condensation

A practical approach for the design professional
[ Page 2 of 3 ]      
Sponsored by Duro-Last®, Inc.
Dr. James Hoff

Roof Condensation and Roof Leakage Look the Same

The observation of liquid water within a roofing system does not necessarily indicate that vapor condensation has occurred. Moisture may also enter the roofing system via leakage through cuts, punctures, loose seams, and many other discontinuities in the exterior waterproofing layer of the roof. And because roof leakage continues to be a widely reported phenomenon sufficient to fund thousands of roof service operations across the country, excessive moisture in a roof may be attributed more frequently to exterior leakage rather than interior condensation. In fact, several of the research papers reviewed further in this article include observations of roof leaks that may better explain the presence of moisture for some observations of what initially may have been classified as condensation.

Vapor Retarders Help, but There are Unintended Consequences

Installing an impermeable material as a vapor retarder beneath the dew point in a roof certainly will address many of our concerns regarding condensation, but the protection of a vapor retarder comes at a price. If a roof with a vapor retarder experiences roof leaks, the water entering the roof will be trapped above the vapor retarder within the roofing system. However, because the vapor retarder will prevent the water from entering the building, the leak may be unobserved and left unattended for a considerable time. And during that time, the trapped water may spread for a considerable distance from the point of the leak, making it even more difficult to pinpoint the source of the leak and causing extensive damage to a constantly growing portion of the roof. Figure 3 illustrates how a vapor retarder can reduce the potential for vapor condensation but also increase the potential for trapped water due to roof leaks.Figure 3

Many Roofs Will “Self-Dry” If We Let Them

The self-drying roof concept may be traced to field accounts of roof systems observed to have been wet in the winter but dry by the following summer. This condition of wetting in the winter and drying in the summer has now been documented in numerous research studies (see Desjarlais et al., 1998, and Griffin & Fricklas, 2006, Chap. 6). And a considerable body of knowledge has been developed to understand how the phenomenon operates. Self-drying roof systems use the sun and summer heat to evaporate moisture that may have accumulated in a roof due to condensation over the winter. The external heating of the roof reverses the vapor drive and transfers the condensed moisture back into the building where it originated.

The self-drying roof is not only an accidental discovery but also a sometimes- overlooked benefit for the roofing industry. Although under some conditions roofs may experience moisture condensation in the winter, the dynamics of summer heat and sun help to mitigate any potential long-term damage from this winter condensation. However, the self-drying process may only be effective for roofs installed over vapor-permeable decks such as metal or wood and only if a vapor retarder has not been installed as part of the roofing system.

Predicting Roof Condensation: From Rules of Thumb to Computer Models

Due to the effects of the perm rating of materials, the potential for condensation- related damage, and the inadvertent movement of air within a roofing system, the initial scientific simplicity of diffusion- driven vapor movement becomes significantly more complicated—and risky. And the adverse implications of underestimating the potential for condensation within roofing systems may influence roofing professionals to take a conservative approach when assessing the impact of these additional factors. At the same time, addressing the risks of condensation by installing a vapor retarder may result in increased damage from roof leaks and eliminate the self-drying potential of the roof. Because of these risks and consequences, building researchers and roofing professionals have expended considerable effort to develop useful tools to aid in the analysis of condensation in roofs and select the best designs to address the effects of condensation.

One of the earliest approaches to roof condensation was a simple rule-ofthumb: If the roof is located in a climate with an average January temperature of 40 degree Fahrenheit or less, condensation should be assumed and a vapor retarder should be installed. A more sophisticated approach was developed by Tobiasson (1994) who identified the minimum average January temperature that would require a vapor retarder based on the interior humidity of a building. Figure 4 provides two U.S. maps to illustrate these two approaches.

Source: Oak Ridge National Laboratories

Figure 4: Simplified approaches for recommending a roof vapor retarder

The map on the left illustrates the geographic extent of the traditional 40 degrees Fahrenheit rule-of-thumb, while the map on the right illustrates the maximum interior humidity allowed before a vapor retarder is recommended.

As illustrated by the two maps in Figure 4, the 40 degree Fahrenheit January temperature cut-off is more conservative than Tobiasson’s indoor humidity approach. As an example, the 40 degrees Fahrenheit map locates Cincinnati, Ohio, within the area requiring a roof vapor retarder. However, the indoor humidity approach would recommend a roof vapor retarder only if the indoor humidity in January exceeds 50 percent RH for a roof in Cincinnati. This is an important difference because many building occupancies, from residences to stores and warehouses, will rarely develop an average humidity above 50 percent in the winter.

Recently a more sophisticated approach to predicting roof condensation has been developed. This approach is called WUFI and was the result of a cooperative effort between cooperation with Fraunhofer Institute for Building Physics in Germany and Oak Ridge National Laboratory in the United States. WUFI is a free software program available from ORNL1, which allows modeling of heat and moisture over time in complex roofs and wall assemblies. WUFI incorporates the newest findings regarding vapor diffusion and liquid transport in building materials and has been validated by comparison with field observations. Because WUFI can provide a day-by-day analysis of moisture movement based on actual weather data, it can be used to plot the characteristics of roof self-drying so that the analysis of condensation may be viewed as a longterm phenomenon. WUFI also allows the inclusion of the effects of solar radiation, which may be an important factor in roof self-drying.

Although WUFI obviously marks a significant step forward in building moisture analysis, it still may be subject to some of the same conservative, rule-of-thumb principles that characterized previous simplified approaches. As an example, the WUFI user must make assumptions regarding air movement within the building; and as discussed previously, the amount of air movement can have a dramatic effect on moisture transfer and the potential for condensation. As a result, it may be very easy to overestimate the amount of air movement and in turn overemphasize the potential for long-term condensation within a roof or other building assembly. In addition to issues regarding air movement levels, the current WUFI tool contains only limited data for popular North American roofing system components such as polyiso foam insulation.

Recent Roof Condensation Research

Because of new tools like WUFI, research into roof condensation has increased dramatically within the last few years. And because WUFI can incorporate solar radiation effects and the solar reflectivity of the roof surface, much of this research has been directed at how moisture movement is affected by a cool, highly reflective roof compared to a darker roof with low solar reflectivity. One of the first research studies to be published (Bludau et al. 2008) introduced two concepts regarding cool roofs. The first involves the reduction in internal heat gain caused by a highly reflective roof during the day, and the second involves nighttime radiation effects that can cool the roof surface below ambient air temperature. Modeling conducted in this study suggested that a white reflective roof installed over a layer of thermal insulation (in this case, polyisocyanurate) and a permeable (i.e. self-drying) metal deck in an extreme cold climate zone like Anchorage, Alaska, may accumulate moisture in the winter in excess of the amount of moisture that can be removed in the summer through self-drying. Modeling conducted on the same roof assembly in a less severe climate (Chicago) suggested that the amount of moisture in the roof system would tend to increase in the winter, but the self-drying process in the summer would still return the roof to its original moisture condition. A similar modeling study (Saber et al., 2011) which included a layer of wood fiber instead of polyisocyanurate foam suggested that excessive moisture accumulation would occur under white reflective roofs in northern locations such as Saskatoon, Saskatchewan and St. Johns, Newfoundland. However, modeling for a similar roof in Toronto showed no longterm moisture accumulation.

In the same time period of these modeling studies, several professional consultants began to investigate field reports of excessive moisture accumulation in cool roofing systems. Hutchinson (2009) summarized a number of anecdotal reports of roof condensation in cool roofs in and around Chicago. In all cases, the observation of condensation was associated with one or more roof design problems such as the use of a single layer of roof insulation or workmanship practices that allowed excessive air movement. In a similar manner, Dregger (2012) reported field observations of three cool roofs installed in California which also showed excessive moisture accumulation. All of these roofs were installed over plywood or OSB decks with no insulation above the deck; and Dregger concluded that the moisture accumulation observed could be attributed to a combination of a lack of thermal insulation above the roof deck and excessive air movement within the roof assembly. Dregger also suggested that the addition of a layer of insulation above the roof deck and under the roof membrane would have corrected this problem by moving the dew point below the roof deck and reducing air flow within the roofing system.

As a result of these initial modeling studies and field observations, a new level of experimental research was started to better understand the potential for excessive moisture accumulation in cool roofs. Ennis and Kehrer (2011) conducted a study of mechanically-attached single-ply roofs installed in Chicago (the same location as Hutchinson’s field observations), and used the results of the observations to conduct a modeling study using WUFI. In all cases, both black and white roofs modeled dried out completely by the summer even if small levels of moisture may have accumulated during the winter. In addition, Ennis and Kehrer observed that the WUFI modeling tended to predict a higher level of moisture accumulation in the white roof than actually observed during field observations. The researchers suggested that the difference between predicted and actual moisture accumulation may be due to a relatively high air exchange rate selected in the modeling to account for possible air movement due to billowing of the mechanically attached roof membrane.

In an effort to better quantify the amount of air intrusion in mechanically attached single-ply systems, the National Research Council of Canada (NRC, 2104) conducted an experimental observation of air movement using a large pressurization apparatus typically used for wind uplift testing of roofing systems. After measuring actual air leakage of various mechanically-attached single-ply systems, the NRC study concluded that leakage was negligible and well below the minimum ASTM and ICC standards for a building air barrier. As a result, the NRC research tends to confirm previous suggestions that the assumed air infiltration rate for mechanically- attached single-ply roofs is much higher than the amount of air infiltration that actually occurs.

Finally, a large-scale field study of cool single-ply roofs was conducted by Fenner et al. (2014). This study looked at cool mechanically attached single-ply roofs installed on 26 Target stores, all in colder climates, all more than 10 years old, all direct to steel deck, and half with only a single layer of insulation. Based on roof test cuts conducted at each store location, visible moisture was observed in only one roof; and this moisture was attributed to an observed external leak rather than internal condensation. A follow- up by this author, conducted using a portable humidity sensor in several Midwest Target stores during December 2014, suggests that the actual humidity in the stores in the winter may be lower than assumed in previous modeling studies. These anecdotal observations indicated that the average humidity in these stores in the winter month of December ranged from 25 percent to under 35 percent during outdoor temperatures from 25 to 30 degrees Fahrenheit.


[ Page 2 of 3 ]      
Originally published in Building Enclosure