Ventilating Attics to Minimize Icings at Eaves Using outdoor air to ventilate the space below a snow-covered steep roof is an effective way to avoid icicles and ice dams.
	by Wayne Tobiasson, James Buska, and Alan Greatorex
	Two identically constructed roofs photographed at the same time. The building with no icings was unheated.

Icicles and ice dams form at the eaves of some roofs in cold regions. Water that ponds behind ice dams may leak into the building since most steep roofs are configured to shed water, not hold back standing water.

The photographs above show two roofs of identically constructed buildings located near Watertown, New York. They were taken within minutes of each other. One roof contains large ice dams and icicles, but the other is ice free. Why? The snow on top of the chimney of one roof is the clue to the difference in behavior. That building was not being heated while the other building was.

This example demonstrates that building heat, not the sun, is the primary cause of ice dams and icicles on roofs. When the sun melts snow on roofs it also warms the eaves, and this tends to minimize the growth of icicles. Certainly some icicles can form on unheated buildings and from solar heating, but they are usually small, infrequent, and do not cause chronic problems.

Research conducted in 1976 at the University of Minnesota concluded that a combination of insulation, ventilation, and "correct house design" is needed to reduce ice dam formation. Recent studies also promote use of "cold" ventilated roofing systems to reduce icings at eaves. These studies also indicate that icings can be reduced by increasing the slope of the roof, by making the surface slippery so that snow slides off, by not installing gutters and by reducing the overhang at the eaves. However, on roofs without gutters, too small an overhang can cause wetting of the walls below or formation of icings on them. A 12-inch overhang is often a good compromise in cold regions. Also, allowing snow to slide off roofs can create hazards. Snow guards may be needed to hold snow on slippery roofs.

A few years ago many office buildings, barracks, and dining facilities were built at Fort Drum near Watertown, New York. All of these buildings have standing seam metal roofing systems above ventilated attics. Some of the roofs remain clear of icicles and ice dams, several experience some problematic icicles and ice dams, and a few experience severe icicles and ice damming. This range in performance is related to the ability of each building's attic ventilation system to remove heat that enters the attic from the warm building below and heat produced by HVAC equipment located in the attic.

We are developing recommendations for solving these specific problems and are attempting to better understand how and when icicles and ice dams form. During our first winter of study (1990-91) we monitored four buildings to determine their current performance before modifications were made to improve the ventilation of their attics. We selected the four buildings to study a range of icing problems from some to severe. By "some" we mean minor icings had occurred along the eaves and large problematic icings had developed at some locations such as the base of valleys. A building with severe icings is shown below. We also monitored a nearby building not experiencing icing problems as a control.

We monitored the outside air temperature in a small weather shelter in the vicinity of these buildings. Attic air temperature was monitored near the middle of each attic. We took temperature measurements once an hour from November into April. Battery-operated data collection systems stored the data between our periodic visits.

Observations by ourselves and others indicate that icings seldom grow when the outside temperature is above 22°F.

Plots of attic air temperature vs. outside air temperature are presented in Figures 1 and 2 for buildings experiencing no icing problems, and "severe" icing problems. Horizontal and vertical lines representing, respectively, an attic air temperature of 30°F and an outside air temperature of 22°F are also presented on each figure. The portion of each graph to the right of the vertical 22°F line is warmer than conditions observed to create icings. The portion of each graph below the 30°F horizontal line is also not within the "icing envelope" because the attic is then so cold that snow on the roof is not melted by building heat.

We chose 30°F for the horizontal line instead of 32°F since we expect that there were places in the attics that were somewhat warmer than the places where our temperature sensors were located.

Of the four quadrants created in Figures 1 and 2 by the 22°F vertical line and the 30°F horizontal line, the upper left quadrant defines the problem area (the "icing envelope").

As we expected, for the roof with no icing problems (Figure 1) very few data points fall within the icing envelope. For the roof with "severe" icing problems (Figure 2 ), 23% of the data (23% of the time during the winter) falls within the "icing envelope."

The separate line of points in Figure 2 that runs down toward the lower left corner of the graph represents a five-day period when the heating system of that building was off due to mechanical problems. Those points provide further evidence that building heat is the primary source of icing problems since, once cool, that building performed out of the "icing envelope."

The curves of best fit for these two roofs are the upper and lower solid lines on Figure 3. The three lines between them are for other buildings we monitored that also had "some" icing problems. All this information suggests that icings can be avoided by sizing attic ventilation systems to maintain an attic temperature of 30°F when the outside temperature is 22°F.

Figure 1. Attic temperature versus outside air temperature for a roof experiencing no icing problems.

Figure 2. Attic temperature versus outside air temperature for a roof experiencing severe icing problems.

Figure 3. Lines of best fit for the roofs shown in Figures 1 (bottom line) and 2 (top line). The three circled lines are for other roofs with "some" icing problems.

Applications

The attic described by Figure 2 experienced severe icing problems. However, it only needed help in the form of improved natural ventilation or mechanical ventilation for less than 23% of the winter. We were not able to provide enough inlet area to completely solve this attic's icing problems using only natural ventilation. Thus several large fans were installed near the ridge. The fans are not dampered. This allows the fan openings to serve as outlets for natural ventilation, thereby reducing the amount of time that mechanical ventilation is needed.

We installed instrumentation to monitor when the fans are needed. The fans are thermostatically controlled, since they are needed infrequently. They operate only when the attic temperature is above 30°F and the outside temperature is below 22°F.

We monitored the modified building and a similar unmodified building from November 1993 into February 1994 (see Figure 4). Both buildings were having similar severe icing problems before one was modified. The portion of each data set to the right of the 22°F outside air temperature line in Figure 4 relates to natural ventilation since the fans cannot operate when it is warmer than 22°F outside. The dramatic difference in that portion of the two data sets indicates that natural ventilation has been improved significantly. We expect that much of this improvement would not have been achieved if the fans contained louvers that were opened only when the fans were on.

The "hunk" taken out of the data set for the building with improved attic ventilation reflects the contribution of the mechanical ventilation system. The mechanical system had been able to keep that attic out of the "icing envelope" most of the time. Without mechanical ventilation it appears that the attic would have operated within the icing envelope for a significant amount of time with problematic icings expected. This verified our feeling that natural ventilation alone would not solve the icing problems being experienced by these buildings.

The pictures, below, show the two buildings on the same date (January 12, 1994). The unmodified building is subjected to severe icings all along its eaves. There are only a few small icicles at the base of the valleys of the building with improved attic ventilation. All other irregularities along the eaves of that roof are snow cornices, not icicles.

Seven large fans were installed to mechanically ventilate the modified attic. Each one consumes about one kilowatt of power. Using the methodology we've described in "Design Approach" (below), four of these fans would be enough to do the job. So, in February we had three of the fans turned off and blocked with sheet metal to preclude both mechanical and natural ventilation through them.

The natural ventilation portion of the data for openings provided by four fans did not change noticeably from that when openings were provided by seven fans. The mechanical ventilation data indicates that the four fans kept the attic out of the "icing envelope" almost as well as the seven fans did. No large icings formed on the modified building with only four of the seven fans working. These findings convince us that the design approach presented in this article can be used to size natural ventilation systems and, quite likely, also mechanical ventilation systems for solving icing problems.

Figure 4. Attic temperature versus outside air temperature for a building with severe icing problems and a similar building with improved attic ventilation experiencing no icing problems.

These two buildings were photographed at the same time. The building shown on the left has its original venting, and is experiencing severe icings. The building on the right with improved attic ventilation has no icing problems at all.

Problematic icings appear to develop very slowly, if at all, when the outside temperature is above 22°F. We feel that owing to variations in temperature within an attic, design should be based on an attic temperature of 30°F. Thus we recommend that to eliminate icing problems, attic ventilation systems be sized to maintain an attic temperature of 30°F when the outside temperature is 22°F. We have modified several buildings to improve attic ventilation using these guidelines. Severe icings did not form on them last winter. Instrumentation we installed to monitor their performance appears to validate our design approach. Properly designed attic ventilation systems that create "cold" ventilated roofs avoid the many problems associated with ice dams and icicles along roof eaves.

Our research was conducted on relatively large buildings. Other work we have done using the same design approach indicates that icing problems on most smaller residential buildings can be solved by providing or improving on natural ventilation (mechanical ventilation is usually not needed). In homes it is important to insure that the natural ventilation provided to cool the roof is not somehow blocked. Also, heating and ventilating ducts that pass through the attic should be well sealed and insulated and the heat they add to the attic should be considered when sizing the ventilation system. Finally, good insulation and continuous air barriers between the living space and the attic are essential so as to minimize the passage of heat and warm air into the attic (see "Beauty and the Beast Upstairs," p. 27). In cold regions, vapor retarders are often necessary to reduce moisture migration. The ventilation provided to minimize icings also serves as a second line of defense against accumulation of moisture in attics.

Wayne Tobiasson and James Buska are research civil engineers, and Alan Greatorex is a civil engineering technician with the U.S. Army Corps of Engineers Cold Regions Research and Engineering Laboratory (CRREL) in Hanover, New Hampshire.