Here's the answer. This past spring we were asked to investigate a "problem job" that was part of a new construction project installed last winter. The problem with this roof was that the polyisocyanurate roof insulation was found to have uniform gaps between boards of 3/4" to 1 inch. Insulation boards measured on the roof were as small as 95 3/8" long and 47 3/16" wide. We also noted that most of the insulation boards were cupped.

We concluded that the cause of this shrinkage was cold weather related. Subsequently, 40,000 square feet of installed insulation was rejected and torn off, and new insulation was installed. 

It is a significant problem when there are board gaps this large in a roof system. One obvious problem is that the roof is not uniformly insulated, and the cost to heat and cool it will be higher than it should be. Board gaps also create thermal shorts in the roof, and may cause moisture to condense in these gaps. Finally, board gaps may create stresses on the roof membrane, and result in splits in the membrane. So board gaps of this size are clearly unacceptable in roof construction.

While it is nice and hot outside now, I know there are winter roofing projects being designed, bid and awarded right now, that will suffer the same fate.

So how did this insulation board shrinkage occur? It was tempting to suspect that this shrinkage was due to the new pentane blowing agent insulation manufacturers recently have converted to. As it turns out, the blowing agent for this insulation was the same HCFC-141b that has been used for several years. So the shrinkage was not caused by a change in the blowing agent.

Was it a manufacturing defect? While insulation boards on the roof were 5/8" short and narrow of the expected 48" by 96" dimensions, our field measurements of insulation taken from job-site storage showed those boards to be 48" by 96" +/- 1/16". Laboratory testing of boards taken from storage showed these boards to have normal cell structure and orientation, whereas boards taken from the roof showed collapse of the closed-cell structure within the board. The boards from the roof also showed edge cavitation (compression), indicating cell collapse.

Here's what we think happened. The roof insulation was installed in December and January on this new construction project. The roof membrane was then installed, and the building was closed in. Then, typical for northern construction practice, the general contractor brought in propane heaters to dry out the frozen ground. This was done so they could pour the concrete floor slabs, which they did beginning in mid-January.

A significant amount of moisture was created by the combustion of propane, the thawing and drying of frozen wet ground and from the hydration of curing concrete. Having closed the building in, a strong vapor drive was created, causing the moisture to condense in the polyisocyanurate insulation. As night-time temperatures dipped below freezing, the water vapor in the cells of the foam caused the cell structure to rupture and collapse, resulting in shrinkage of the board. These two elements then were the catalysts for cell collapse and board shrinkage: high humidity and freezing temperatures.

Polyisocyanurate foam is the best insulator we have in the roofing industry, but it can collapse, as it turns out, like a fallen soufflé' under the sub-freezing temperatures and high humidity conditions. As designers and insulation consumers, we need to understand when these cold damp conditions might exist, and provide a vapor barrier or retarder, even if these conditions may only exist for a short (but critical) duration.

Is this collapse well known? I called all the polyisocyanurate insulation manufacturers, and all of them said this same thing had happened to them - cold weather, pouring of concrete in a closed-in building, cell collapse, and board shrinkage.

The roofing insulation manufacturers are somewhat reluctant to require designers to install a vapor retarder for new winter construction projects. If the other manufacturers don't do it also, then they may miss getting a sale. So, one manufacturer has inserted this disclaimer in its literature:

"Special consideration should be given to construction-generated moisture, as well. For example, construction-generated moisture, such as will be released by placing concrete floor slabs after the roof has been installed, can drive large quantities of moisture into the roof system."

What isn't explained in this disclaimer is that the insulation can shrink and collapse. So if you are planning a new construction project where the roof is scheduled to be installed prior to the concrete floors being poured, and this is to happen during freezing temperatures, it is our recommendation that the roof design include a vapor retarder.