Location of smart vapor barrier in double stud wall
Considering a double 2×4 wall with dense cellulose. Zone 5. To defend against moisture from cold sheathing in winter months, plywood (rather than the less permeable osb) would be used with a rain screen. My understanding is that a smart vapor barrier on the interior could be helpful in keeping moisture from diffusing outward. A smart vapor barrier would also double as an excellent air barrier…until I do anything that penetrates the exterior wall surface and punctures the barrier, then it becomes mediocre at best. Would it be better to put the smart vapor barrier on the inside of the outermost 2×4 wall rather than the inside of the innermost 2×4 wall? It would be more effective as an air barrier, but at what cost?
GBA Detail Library
A collection of one thousand construction details organized by climate and house part
Replies
OML,
The vapour-retarder needs to be close enough to the interior that it stays warm enough not to attract condensation. The rule of thumb used is that 2/3rds of the R-value should be outside it, but different codes have different requirements.
I find that the sheathing makes the best location for the primary air control layer, whether taping the sheathing itself, installing a self-adhering WRB or a mechanically fastened and taped WRB. The interior vapor control layer can also be made airtight, but as you note, it's hard to keep it continuous.
The IRC that most of the US uses as a basis for local codes only specifies that a vapor retarder should be installed "on the interior side of framed walls." (R702.7:https://codes.iccsafe.org/content/IRC2018/chapter-7-wall-covering#IRC2018_Pt03_Ch07_SecR702.7). I design a lot of double-stud walls in CZ6 and always include a variable permeance membrane at the interior side, as highly insulated walls have little heat energy available to push moisture through sheathing so I want to keep as much interior water vapor out as possible. The farthest into the wall assembly I've put the vapor control layer is 1/4 of the R-value and monitoring shows that it works just fine there.