Water dripping from a skylight during an autumn rainstorm looked like a minor leak. But when the roofer making repairs discovered an unexpectedly large area of saturated OSB sheathing and extremely corroded nails and plywood clips, he called me to take a look. We agreed that this was more than a flashing problem.
The roofer and I had worked on the house together 13 years earlier, when it was built. I was the framing contractor; the owners were the GCs. Vaulted ceilings were a key part of the home’s design, and I’d framed much of the 6/12-pitch roof with raised-heel scissor trusses. The owners then had the trusses insulated with open-cell Icynene spray foam, creating an unventilated roof assembly. Icynene was also used to insulate the walls of the single-story home, which was built on a permanent wood foundation surrounded by berms on the west, north, and east sides. Inside, the soil was covered with compacted stone and a layer of poly taped at the seams; the pressure-treated floor joists were installed on top. The home is heated with a natural gas furnace and has central air conditioning and a Venmar vanEE 1000 Duo energy-recovery ventilator (ERV). Supply ductwork in the attic was also insulated with Icynene, while uninsulated return ductwork runs underneath the wood-frame floor.
I wondered if the home’s unusual construction and unvented roof were contributing to the problem, so I called Silas Hoeppner of Cenergy, a local HERS rater and energy consultant, to inspect the foam installation. Hoeppner didn’t discover any major defects, but he did find several small voids, some of which corresponded with areas of damaged sheathing that we found. Those voids could have allowed some moisture to be transported by air movement to the sheathing.
Vapor diffusion — the movement of moisture from areas with high moisture concentrations to areas with lower moisture concentrations — was also a concern. Icynene is a good air barrier, but unlike closed-cell spray foam, it’s vapor permeable. The IRC requires a vapor retarder coating to slow diffusion when an unvented roof is insulated with open-cell foam in climate zone 5, where this house is located. While this installation had no vapor retarder — and therefore did not meet code — Hoeppner noted that the manufacturer’s specs don’t recommend one in Des Moines’ 6,600 heating-degree-day climate, since it could interfere with drying to the interior during warm months, when the air conditioning is operating and the interior of the house is cooler and drier than the exterior.
Most of the roof deck was sprayed with 9 inches of Icynene, which has a stated R-value of 3.6 per inch. But the insulation was sprayed unevenly, and Hoeppner measured depths as little as 6 inches (R-21.6). Even assuming an average R-value of around 30, the roof had less insulation than the IRC’s current R-38 minimum for zone 5.
Measuring Humidity and Radon
To monitor the home’s temperature and humidity levels, I installed a couple of simple UEi THL1 data loggers (877/571-7901, tequipment.net). I mounted one high up on one of the vaulted ceilings in the living space, and another in an attic area. Using software that came with the data loggers, I charted the data over a period of two winter weeks in December 2009 and found that the home had relative humidity (RH) levels that ranged from 40 to 48 percent in the conditioned attic but rose as high as 56 percent in the living space, even with the ERV operating.
Given the home’s unusual earth-bermed construction and the high radon levels in our area, radon testing also seemed like a good idea. As it turned out, the home had levels exceeding 38 picocuries per liter (pCi/L), much higher than the EPA’s 4 pCi/L action level. So my clients installed a radon-mitigation system — basically a continuous-duty fan that depressurizes the subsoil under the ground-level vapor barrier and exhausts radon-containing soil gas to the outside. With the new radon system installed, radon levels dropped below 3 pCi/L later that winter. As an added benefit, the installer reported that RH levels in the house had dropped beneath 30 percent, a more normal wintertime level.
Depressurizing the soil under a house also tends to dry the soil out, so I wasn’t surprised by the drop in indoor humidity. I was concerned that the roof damage had been caused by excess water vapor from the soil migrating through the spray foam — either through diffusion or, more likely, air leakage — and condensing on the cold sheathing. But I couldn’t rule out the possibility that the wet sheathing was the result of ice dams caused by conductive heat loss, especially considering the modest levels of insulation and the roof’s relatively low pitch.
With indoor humidity now in check, I was less concerned about vapor migrating through the Icynene and condensing on the roof sheathing. Still, there wasn’t enough insulation in the roof, so I recommended stripping the entire roof and adding a continuous layer of 2-inch-thick extruded polystyrene (XPS) on top of the roof deck, then installing 2x4 sleepers, new sheathing, and new shingles. The new foam and sheathing would more than triple the cost of reroofing the house, but it was the least expensive way to add needed insulation to the vaulted ceilings. Owens Corning Foamular 250 has an R-value of 5 per inch, so 2 inches of foam would bring the R-value of the roof assembly up to about R-40. The continuous layer of foam would also keep the OSB sheathing warmer, so that it would be less likely to become a condensing surface.
Since we would have to install sleepers over the foam to attach new sheathing, I planned to take advantage of this by also using the sleepers to provide ventilation to the new roof deck. This would help keep that sheathing cold and prevent the formation of ice dams. Of course, adding extra thickness to the roof meant that we would also have to address the existing fascia and gutters and remount the skylights.