Ice Dams: Causes, Contributing Factors, Diagnosis, Management, Expectations

By Jon Haehnel and Timothy O’Dell "I just don't get it. I just re-roofed and added insulation three years ago, and now the problem is back," the exasperated manager complained as he showed me his water-stained second floor ceiling. At least he had the "cold comfort" of not being alone, I thought. The winter of 2010 - 2011 was a good year for ice dams, allowing an uncommon use of the word "good." Roofs that had not had icing problems for years carried ice a foot or more thick for weeks on end. Shovelers and roof-hoers stayed busy. Others struggled with yearly problems on roofs that always iced-up, but were especially troubling this year. Roofers' phones rang off the hook as building owners and managers rushed to stop spreading damage and to prevent further episodes.

Roof leaks, risk of injury from falling ice, or the specter of an attic grown thick with mold focus attention on stopping leaks and removing the ice. With the emergency over, some ask why this year? Why my roof and not my neighbor's? Why only those places? What can I do?

We offer answers to such questions here, and we hope, useful perspective based on our fifteen years combined experience with roof icing in the far northeastern U.S. In our view, virtually any roof will have episodes of icing given the right combination of factors. Below we discuss factors contributing to roof icing that we have come to recognize. We argue for management measures for specific conditions, as opposed to general solutions, to this all-too-common fact of cold climate life, including occasionally doing little or nothing. The answer to "What is the problem with ice dams" varies with owners, by year, by building and by location on the same building. Reasonable expectations for solutions vary. Given ranges of variability, and sometimes complex interactions between many factors, we advise giving or taking guarantees with caution.

In general, ice dams form when melt water shedding down slope encounters a roof surface where the temperature is 32F (0C) or lower. Formation is typically, though not always, at overhanging eaves, where temperature of the roof surface is likely to be sharply colder than at areas above heated space. Water holds to the roof edge temporarily by a property called adhesion and before it can get a final shove by gravity it freezes in place. With the first drop frozen in place subsequent drops of water build up behind the first and freeze in place and a dam begins to form and spread along the eave. In principle, icing can happen anywhere on a roof where the surface is cold enough to freeze melt water before it can flow any further. As an example, consider growth of an icicle on which water adds to the end before it can drip to the ground.

A related surface effect is the way in which snow absorbs melt water. New snow cover has a high internal surface area that acts like a sponge, soaking up melt water rather than allowing it to shed. When the temperature inside the water soaked "sponge" falls to freezing, it becomes the start of an ice dam.

Noting the occurrence of "surface" in each sentence of the preceding two paragraphs, and its association with critical temperature and speed of shedding, one grasps the dynamic nature of the icing process and the way roof ice grows and shrinks depending on the balance of the three. Whenever a critical mix of conditions is present at the roof surface, icing will occur and increase. With this in mind, one is ready to consider the broad range of contributing factors affecting roof icing.

Contributing factors:
Depth and character of snow:

Thermal resistance of snow, its capacity to restrict heat movement, varies between R-.25 and R-1.1 per inch of depth. Light, dry, "powder" snows have high R-values because they have the high "loft," with many, finely dispersed air spaces. The principle is the same as for loose-fill insulation in a garment, in a wall, or in a roof. Clearly, deep snow cover can provide temporary insulation value above a roof surface that is greater than that installed between the roof surface and heated space below. Under such a condition, roof surfaces will gain heat. With respect to icing, the function of roof or attic floor insulation is to hold the roof surface at or below 32 F. Under sufficiently deep, newly fallen snow on a heated building, the 32 F freeze - melt point will commonly be in a plane above the roof surface and in the snow layer. All snow between this freeze-melt plane and the roof surface will begin to melt and run down the roof.

The location of the freeze melt plane is calculable given ambient temperature below the roof, the total R-value of the roof assembly, the temperature outside, the snow depth and an estimate of the snow's R-value. All is well so long as the 32 F temperature plane is at or below the roof surface. After a few rounds of such calculations, however, one concludes that at many snow depths commonly seen in our region, no reasonable amount of insulation in roof or attic will keep the roof surface cold enough to prevent melting under ALL snow conditions and costs to insulate to prevent icing under all snow conditions are prohibitive.

Effectiveness of ventilation below the roof:
Because sufficient heat will be trapped in attics and under roofs to produce roof icing from time to time, the usual expedient is to exhaust the heat by passive and sometimes active ventilation. The U.S. Army Cold Regions Research and Engineering Laboratory (CRREL), Hanover, NH has done leading research on the subject. They collected attic and exterior ambient temperatures for identical buildings (barracks, dining halls, offices) at Fort Drum, NY, some with substantial roof icing problems and some with none. In these "controlled experiments," CRREL identified two robust icing parameters: They state with confidence that roof icing is likely when ventilation is insufficient to hold attic temperatures below 30F when outdoor ambient is 22F or lower. They go on to present formulas to engineer vents large enough to avoid roof icing conditions.

Ridge vents, soffit vents, gable vents, vent chutes and other schemes to release warmed air keep roof surfaces cold provided that they exhaust heat faster than heat is gained. Clearly, ridge vents do not work when covered with snow. Gable vents must have enough free area and if screened, must not be clogged. Venting of slow pitched roofs, as at shed dormers, 4-pitch ranch homes, gambrels, and at slope-to-wall terminations is either not present or notoriously ineffective. Nevertheless, all venting that retards warming of attic spaces and roof assemblies, whether by heat loss from below or by solar gain, retards snow melt and thus refreezing at the eaves. Effective ventilation helps.

Internal losses - effectiveness of insulation and air sealing:
In days before extensive insulation, when coal or wood fires burned virtually ¨24/ 7¨ roof icing was for many buildings not the concern that it is today. Heat loss through roofs melted snow quickly and continuously, and steep pitches shed it quickly. We still see the occasional under-insulated building where this happens. No one today would choose this as a means to prevent roof icing. The common case today is an insulated roof that receives enough warmth from below to alternate between freezing and thawing. Optimally, a cold attic or the air channels of the vented cold roof would be close to outdoor ambient to retard melting. Therefore, unintended warming of attics and roof assemblies by unintended heat conduction and unintended warm air movement from other parts of the building is to be minimized. Conditioned spaces need effective air and thermal barriers both to retard icing and for reasons of economy.

A common misconception is that one can mitigate ice damming by adding additional insulation directly below the dam area. Such a measure fails to consider melt water from far above and often makes the dam condition worse. Why? Because the additional insulation can make the exterior roof surface colder, effectively it makes the refreezing zone larger than just the eaves.

An infrared image of a snow covered roof. The warm location indicated at ridge vent is where snow is melting and streaming down the slope.

Weather patterns:
Ice dam severity varies with the number of days on which conditions for ice damming for a particular roof are met. Deep, light, "powder" snows, followed by several days of bitter cold weather favor the onset on icing on any roof. Some years will be worse than others.

Roof pitch:
Snow that slides off the roof and melt water that drips off quickly before it can refreeze eliminates some of the key factors that create ice dams. The steeper the roof slope the better. For this reason metal roofs are often seen as a "solution" to ice dams because they promote snow sliding but there are important cautions to consider before you switch to a new metal roof. Look at the complexity of your roof, does it have valleys, dormers, or skylights that could prevent the snow from sliding off easily? If a roof valley is narrow the snow can get bound up as both sides of the valley try to slide snow into the same place. Standing seam metal roofing can exacerbate this affect by making chutes that trap the snow in valleys. Snow jams can lead to ice dams just as easily as water refreezing at the eaves. Also, a metal roof is colder than a shingle roof so refreezing melt water at the eaves can still be a problem. The solution: seamless valleys made with rubber membrane (EDPM or TPO). These commercial roofing materials are essentially seamless, slippery, and waterproof. The downside is it can look rather ugly compared to traditional residential roof materials. Metal roofing works best if you have relatively uninterrupted slopes with wide valleys that separate the standing seams.

A typical ice dam - some causes are revealed in the image below.

The contributing factors to this ice dam are the skylight, 3 roof areas shedding into the same valley and the standing seams trapping snow in the valley.

Roof material:
Modified bitumen underlayments, such as Ice & Water Shield, are touted as ¨insurance¨ against roof leaks. In most roofing applications, these "self-healing" membranes are pierced everywhere a roofing nail is driven. A very high percentage, but not all, of these piercings heal. Metal roofing sheds snow faster than any other roofing material but folded and crimped standing seams are not waterproof. These products and all other commonly used residential roofing provides water shed but are not water proofing. Ponded water is very ¨patient.¨ If there is an opening below, the water will find it. Both bitumen underlayments and metal roofing are good materials but we have seen all of them fail at least once to provide 100% protection against roof leakage from water ponded behind ice dams.

As mentioned above, rubber membrane roofing IS waterproof and it is suitable for all roofs, but its use tends to be limited to flat roofs and to commercial, industrial, and institutional buildings.

Roof configuration and under-roof spaces:
With roofs the simple case of two planes of equal slope meeting at a horizontal center ridge is seldom the case. Shedding from pitches above, from two sides of a valley, between closely spaced dormers, from steep pitches onto lower pitches, around skylights, and endless combinations of these and other details all contribute to snow depth and density along lines and at points. Snow shed and melt water concentrated in valleys or on low adjoining porch roofs, for example, can cause localized damming conditions after moderate snowfalls which leave other, more ice-prone roofs unaffected. The case of multiple sheds to a single outlet or a much slower outlet are effective designs for what we like to call ¨ice dam machines.¨

A roof cross-section of our region's ubiquitous cape-style home provides a useful illustration of the effect of under-roof spaces. From eave to ridge, these homes commonly have overhanging eaves, unheated kneewall spaces, sloped ceilings installed directly on rafters, and small, triangular attics above an expanse of flat ceiling. Each portion of this roof section can contribute to icing. The eaves and roof areas above kneewall spaces are typically where ice forms and builds, often starting as snow and water ¨sponge¨ described above. Sloped ceilings with limited framing space for insulation, often poorly vented, are usually first to provide melt water on account of heat conducted from conditioned space below. Snow on slopes above the attic will melt later provided that the attic is not warmed by losses from the interior.

As an example of perverse behavior of such an "icing system," we have seen cape roofs cleared of snow up to the level of the attic floor, which later iced at the eaves from melt above the attic. Solar gain at the cleared portion warmed the roof assembly, which warmed the attic by convection, melted the remaining snow, which froze at the eave.

The valley between these cold porch roofs is a potential "ice dam generator". Melt water from higher up the slope hits these cooler roof areas and refreezes before it can drip off. Note the frost pattern indicates these roof areas are few degrees cooler than the roof further up.

Orientation and time of season:
Solar gain usually helps, but it can also harm. Consider a south east roof pitch that receives sun early in a winter day and is in shade by early afternoon. At mid-winter daytime temperatures, morning melt water can be the start of an ice dam here while southern and western pitches are still dripping at their eaves. One thinks of south facing pitches melting and shedding and north slopes holding snow and forming dams driven by internal heat losses. In fact, in midwinter with the sun in perigee, just the opposite can happen. A north slope may stay uniformly below freezing, holding snow until it blows off or dries directly from snow to vapor. Meanwhile, melt and freeze conditions at the opposite side cause icing problems.

Another example: a house is backed up to mountain on the west side and unobstructed on the east side. The house has a simple "A" frame roof over a vented attic with slopes facing east and west. The attic gets warmed by morning sun the east side and melts snow on both the east and west slopes. The east slope is in full sun so whole roof is the same temperature and the water drips off at the eave. The west slope is in shade so the eaves are cooler than the roof over the attic and the melt water refreezes at the eave. The ice forming on the west side never gets a chance to melt because it never sees the sun. Where roof icing is concerned, we advise taking nothing for granted.

Other Contributors:
In our experience, ice dams can start from small beginnings, but once started, grow until conditions change significantly. The following are seemingly small details that we have seen cause the onset of roof icing:
-Thermal bridging through rafters, which have lower R-value than insulated areas between.
-Recessed light fixtures in sloped ceilings which cause ¨hot spots¨ immediately above them.
-Sky lights
-Shading by trees, other buildings or other parts of the same building for even a few hours. Snow melt from the sun gets into the shade and refreezes.
-Exhaust fans, dryer vents. Often a dam forms in the vicinity of a bathroom vent or dryer vent because these devices melt snow. They don't necessarily have to be on the roof to cause problems either. A vent right under an eave can melt snow too.

The owner wondered why his recently foam insulated roof was still creating ice dams in this particular valley. After clearing some snow and ice off the roof he found the bathroom fan vent cap in a cavern under the snow.

This is an infrared image of chimney melting snow around it. The melt water refreezes at the eave just below the chimney.

With so many factors contributing to roof icing, we never rush to a statement of cause. Owners will call needing to take immediate action against massive ice build ups and water entering, a job for shovels, roof rakes and the occasional axe. In order to take account of all possible factors, thorough investigation of conditions inside and out is called for.

We take exterior pictures of melt patterns both close up and at some distance from the building for complete context. When possible, we view the roof from above after snow has been cleared to understand all ¨sources and courses¨ of melt water. This often means being present as snow is removed or on several occasions while it melts off.

We very often find it helpful to wait for another light snow or frost to reveal melt patterns and show us where the roof is warmest. We look at all exposures. The best time for this is often between first light and just after sunrise.

On larger buildings or on sites with several buildings with common details, we look for ¨controlled experiments.¨ We look for identical details of dormers, valleys, roof intersections and the like, both with and without ice buildup, asking which contributing factors are present. Such comparisons can be most revealing, suggesting effective remedial measures, their reasonable limits, and reasonable expectations going forward.

It is at this point that we order a full energy audit of the building shell, including depressurized leakage measurement and detection, infrared scans, and visual inspection to identify unintended heat transfers to the roof.

If we have learned anything it is to take time, to gather observations and data, and to not rush to judgment. There is no substitute for time on site. The higher the expectation for a solution, the more closely we study the conditions.

Management and expectations:
The first counsel that we give to building owners and managers in our region is that, as long as there are snowy, cold winters and heated buildings, there will be varying degrees of roof icing. What one cannot entirely eliminate one must manage. If we have communicated anything above, we hope that it is that some icing conditions are not 100% preventable by passive means at any cost. Therefore, the question becomes, ¨What is the problem to be managed?" Here is a list of ways that we've heard owners define their roof icing problems. One, some or all might apply in a particular building.
-Water penetration to occupied space with damage to interior finishes. This is most common. Many are content to let roofs ice up as long as the interior is dry.
-Heat loss that drives melting, even though the roof does not leak.
-Personal safety and the risk of injury from falling ice.
-Liability for injury to others from falling ice.
-Damage control by those who have lost gutters and drip edges or had roofs damaged by repeated ice removal.
Owners' perceptions of the scale of the problem and its incidence should suggest scaled responses. The principles that we follow in suggesting remedial measures include the following:

Not all appropriate solutions are ¨passive¨ involving changes to roofing, insulation, or ventilation. Active measures involving powered devices or labor have their place. Electric roof heating cables or heated panels operated as need arises to keep recurrent icing under control are entirely reasonable. We know of one hotel which installed these to prevent any icing of its mansard roof. Any risk of injury to pedestrians below was unacceptable.

Snow removal is the time-honored way to prevent icing. An hour with a roof rake, a few hundred, or a few thousand dollars in the case of a large building, to remove snow from a five to ten year event is a reasonable alternative to a shell improvement costing thousands, or hundreds of thousands, merely as a means to mitigate icing.

Where extensive building shell improvements are selected, they likely need to be justified also for their energy savings. For example, we occasionally find a basement-to-attic chase moving warm air and providing ¨shirt-sleeve¨ warmth in an otherwise cold attic. With the chase closed, roof icing decreases substantially. We have seen insulation of a portion of a roof above an icing location make icing more tolerable. The far more typical case is of an entire roof or attic floor needing insulation and air sealing in order to address an icing condition comprehensively. This often involves disturbances of interior ceilings for access. Dramatic results often require extensive measures.

Don't rely on shell thermal improvements to control a problem that needs a more robust roof. Conversely, don´t expect a roof repair to stand up to overwhelming snow conditions. Don´t rely on either to control a condition driven by building design, orientation or detailing.

Never say never. A 200-year snowfall can happen in most any year when all bets will be off. We advise measures to mitigate roof icing, with consistently satisfactory results. Most ¨solutions¨ can be overwhelmed by extraordinary weather events, however. Some locations on some roofs will always have icing conditions when there are sufficient contributing factors.

There are no absolute guarantees; there is only ¨better.¨ We provide no absolute guarantees and we hope that discussion above is a caution to others against offering or accepting them. The good news is that there is a lot of ¨better¨ to be had.

Return to Home page from Ice Dams