Back to Case Studies

Part 1 ] Part 2 ] [ Part 3 ] Conclusion ]

Case Study

Roof Ice Dam Remediation for Northeast Ski-Area Condominiums

Part III

by Henri C. Fennell
President, H. C. Fennell, Inc. & FOAM-TECH
North Thetford, Vermont, USA.

Phase II Diagnostics and Repairs

Phase II non-destructive diagnostic information was collected as in Phase I. The results showed more detailed information about the locations and severity of the remaining heat losses as the overall heat flow had been significantly reduced. Specific melt spots in the snow cover were now visible and could be associated with interior construction details. Due to the improvement from the first round of repairs, the flow of melt water had been dramatically diminished and was concentrated below the areas of remaining heat loss. As a result of monitoring the depth and location of ice dam formations early in the season, the ice pattern was used as an indicator of the volume of melt water that was coming from each problem area above (This was useful in prioritizing the cost effectiveness of Phase II repairs). Later in the season, the ice depth evened out along the length of the eaves (except at the valleys, which still had the largest ice accumulation), due to the melt water spreading horizontally as it was held back behind the earlier ice formations.

Cathedral Roof Slope Warming – Phase II Problems and repairs

Phase II testing was completed late in the winter of 1997/98. The cathedral roof slopes appeared to be clear of problems except for two areas - below the skylights and along the party walls. Based on the infrared and melt pattern analysis, the following assessments were made:

  1. Wind washing in the lower slope was eliminated except in minor spot locations at inaccessible soffit areas. No additional work was performed in these areas.

  2. Infiltration and heat loss at the zero-clearance fireplace boxes was significantly reduced. The complexity of this triangular ornamental box and the limited work access made the work difficult at best. Some infiltration was still evident where the front face of the box met the two outside walls; however, the amount of warm air flow out of the top of the box into the roof slope vents was virtually eliminated. No additional work was performed in these areas.

  3. Infiltration of cold outside air along the wall to roof intersection of the cathedral gable walls had been eliminated. In the two middle units, these walls form the party walls with the two end units. Even after air sealing procedures, the narrow strip of roof over the party walls remained as one of the two most serious melt areas, in part because they are directly above the valleys. With the wall to roof intersection sealed, it is assumed that the heat source causing this melt pattern was heat coming up from the wall cavities rather than directly from inside the room itself. The air barrier in the cathedral ceilings did cross over the party walls which have double 2x4 walls with a space between them (See Detail / Photo Essay #11). As these bays are inaccessible from the interior, and opening the roof was deemed to be too expensive, the decision was made to block the flow of air from the walls into the attic by sealing the roof bay at both ends. This did not preclude the option of opening the roof at a later date, and it promised to limit the heat reaching the roof and attic by cutting off the stack effect from the double wall and creating a dead air space rather than a constant warm flow. As the bottoms of these roof bays had been sealed when the eaves were blocked and sealed, this only required sealing the readily accessible top of the vent space from the attic.

  4. The fan/light in the guest bathroom showed no infiltration on the infrared scan and the melt area in the roof had been significantly reduced considering that the duct was still routed through the roof bay. While totally relocating this fan would have been the optimal solution, the degree of improvement achieved without major renovations was acceptable.

  5. The repairs in the knee wall attics were performing well. Both the infrared and melt patterns showed significantly less warming at cold temperatures. The remaining warming, visible only at near-freezing temperatures, was believed to be caused by the recessed entry porch light.

  6. Air infiltration around the skylights was virtually eliminated; however, ice formation was noted immediately below the skylights. As the infrared and melt pattern testing did not indicate roof warming, and similar construction on both sides of the skylights had not developed ice, it was assumed that the source of melt water in this specific area was from the skylight itself. A Phase II test was developed utilizing a piece of double-glazed tempered glass installed over one of the existing inoperable skylights. If this test alleviates this problem, better insulating glass or an outside storm panel will be proposed. This unit and the unmodified skylight on the next unit were monitored to determine the following:

    • Would additional glazing R-value prevent ice formation below the skylights?

    • Was the melt caused by the glazing only, or by the curb and framing losses around the skylight?

The result of this test was inconclusive. Late-winter conditions did not produce enough snow to effectively compare the volume of ice produced below the modified and unmodified units. Photos of earlier snow cover indicated that the snow pack went clear up to the skylight curbs. This indicated that the heat loss through the skylight curbs was not a significant contributor to the ice build-up below them.

Attic Roof Warming – Phase II Problems and Repairs

The main attic roofs had improved, but not as much as was expected, given the magnitude of the initial improvements. Most of the attic roof now held snow, the exceptions being the cricket attic areas where the roofs of two units overlapped and above the middle unit party wall. The small isolated attics (cricket attics) just over the party walls between the middle and end units (See detail photo essay #8) were especially problematic as this area was directly above the valleys which concentrated the volume of melt water and ice in one spot. This was a good example of the need to plan for two rounds of testing and repairs. The small cricket area had not originally appeared as a unique problem from the rest of the attic roofs when the Phase I testing (infrared and melt pattern analysis) was performed. Only after correcting the main attic heat loss problems could the crickets be identified as unique problem areas. While these attics were relatively small, they still provided conditions conducive to ice dam formation as well as roof leakage. As a result of the second round of tests in the attics, the following "fine-tuning" details were identified and addressed:

  1. The flat cap areas were performing well over the majority of the large ceiling surface area. The contractor had elected to fill large gaps in the original insulation, add batts over fixtures and bathroom fan ducts, and to use an air barrier material over the batts for convection and ex-filtration control. This was successful in improving the performance of the ceiling insulation except around the edges. At the party walls and rafters the air barrier material stopped at the face of the open framing. It was clear that the air barrier needed to be sealed to the wall and roof sheathing to work as a secondary air barrier to stop ex-filtration from ceiling penetrations. Sprayed polyurethane foam was used to make these complex connections around the rafters, vent chutes, and double party wall stud framing. Cut outs around plumbing and chimney penetrations were also sealed (see Detail/Photo Essay #2).

  2. Infrared testing showed that there was still heat flowing into the attic from the top of the cathedral slopes. This was coming from the vent space in the roof bays over the party walls, the narrow spaces between the doubled skylight rafters, and at the boxed-down I-beam at the slope to ceiling transition. While the beam had been air sealed from below, there was little or no insulation over the steel. Because the batt insulation had originally been installed from below, the air barrier and insulation did not pass over the beam. To deal with the reduced space between the top of the beam and the vent chute, the available space was filled with polyurethane foam to provide an air barrier and an adequate R-value. This foam seal was carried over to meet the air barrier material that covered the attic ceiling batt insulation. The foam was also used to seal the tops of the narrow spaces between the doubled skylight rafters and the tops of the party wall bays (see item #4. below).

  3. The infrared scans showed that the backs of two dropped ceiling area walls were still cold. Further investigation determined that the outside of these walls were unsheathed and open to a vented soffit above the windows. Some of the batts had fallen out of these bays. This area communicated with the attic above. The walls were retrofitted with sheathing and sealed as in the other dropped-ceiling areas.

  4. The tops of the cathedral slope bays over the party walls were sealed. These three bays had been the largest sources of heat in the attics after the Phase I repairs. The bays had batts in them, but no vent chutes or interior sheathing. They were open over their entire length into the walls below. As a part of the foam contractor’s work, these bays were filled as far down the slopes as possible. The two party wall roof bays were the primary source of the warming in the small cricket attics.

Part 1 ] Part 2 ] [ Part 3 ] Conclusion ]

Back to Case Studies