Which Plane Is the Air Barrier in Each Element

The air barrier record describes the barrier as the connected set of surfaces a designer nominates to resist bulk air movement, together with everything that joins those surfaces to one another, and says that on a section it should be possible to trace that line around the enclosure without lifting the pen, with the same holding on plan.

It also removes the most common confusion in one move. An air barrier resists air moving through gaps; a vapour control layer resists moisture diffusing through material. A fully taped breather membrane is an air barrier while remaining open to vapour, and a vapour control layer left loose-lapped and unsealed at its edges is not an air barrier at all, however high its resistance to diffusion.

In a mass timber wall the panel is commonly nominated, and the record says that once that decision is made the taped panel joints are the barrier's laps and every screw, bracket, socket and pipe that enters the panel is a penetration of it. In a built-up twin-skin metal roof and a sheeted metal wall the liner sheet is the air and vapour control element as well as the visible internal face, so its side laps, end laps and perimeter sealing are the airtightness of the building.

In a timber frame wall the nominated layer is a sealed plane fixed across the studs with a service void and a lining in front of it. In a cold deck flat roof the ceiling plane is the only side of the void that faces the building, which the record says makes its continuity a roof decision rather than a ceiling finish. In a masonry cavity wall the inner leaf is the surface linings, services and finishes are fixed to, so it is usually where the line sits. In a composite panel roof the panel closes the air path within itself, which is why the joints between panels carry that duty entirely.

Because the nominated plane is a different kind of thing in each case, so is the way it is lost. A masonry surface is continuous until something is chased into it. A sheet is continuous only where it is lapped and sealed. A structural panel is continuous until it is drilled. A curtain wall's line is a chain of gaskets and joints in which, as the record says, the weakest joint rather than the average one sets what the facade does.

The metal records give the sharpest example of a failure that does not present as one. A liner that looks faultless from below and is open at its side laps has failed at the duty that matters, and the symptom appears later and inside the build-up as condensation on the cold outer sheet, running to a lap somewhere unrelated and reading as rain.

The record says changeovers govern the whole. Where the plane passes from a sheathing board into a roof membrane, the two materials move differently, present different surfaces and usually belong to different installers, and if the detail does not say which laps onto which and what bonds them, the line simply stops at the changeover even though both planes are individually complete.

The transition record describes the mechanism. Nothing in a transition bonds to itself: the band adheres to a primer, the primer adheres to a substrate, and that substrate was prepared by a trade which has usually moved on. Where a plane changes direction the loading on the bond changes with it, from shear to peel, and peel is what opens bonds, which is why a clamped termination and an adhered band do different work and neither substitutes for the other.

The transition record names the recurring failure without euphemism: it is not a poor band but an unowned one, because each side's scope runs up to the junction and stops there, and the junction belongs to neither of them. It asks which package the transition itself falls inside, as distinct from the planes on either side.

The air barrier record adds the places this happens most. The boundaries between work packages, frame to roof, wall to floor, structure to opening, are where the line is most often left unowned, and naming an owner for each is described as a project management act as much as a technical one. At the eaves in particular, where the crossing is left to whichever trade arrives last, it is frequently the point at which the line is lost.

Services and the barrier compete for the same plane. Every socket enclosure, waste pipe, extract duct and cable route that crosses the line turns a designed surface into a set of holes, and if the barrier is drawn only in section then the services layout decides where those holes fall. Putting a service void inboard of the line changes the interaction, because most terminations then stop in the void and only services that must pass outward reach the barrier.

The wall records say the same from their own side. The timber frame record describes deleting the service void to gain floor area as quietly transferring the airtightness of the whole wall onto workmanship at each fitting, and the mass timber record describes the service zone as structural to the concept rather than optional, because every fixing driven into a nominated panel is a penetration of the building's air control line.

Because performance belongs to the connections rather than to the components, the line is normally verified by pressurisation testing of the finished enclosure rather than by inspecting materials. The transition record then states the limitation directly: such a test reads the envelope as a whole and cannot say which junction is leaking or whether a bond will last.

That is why both records treat the record made before covering as the useful artefact. A drawn and photographed record of the completed line, made before it is covered, is what makes a later fault findable, and without it the search for leakage runs over the whole enclosure instead of over the part that was disturbed.

The record ends with the part that outlasts the project. Later work crosses the line: a new socket, an extract fan, a loft hatch, a rooflight, a rewire. Whether the household and its trades know the line is there determines whether it is reinstated after each of them.

The internal insulation record makes the same point about a lining that looks like an ordinary wall, and says a drawing of what was installed, where the control layer runs and where fixings may safely be made is what gives ordinary later work any chance of missing it, adding that where that drawing is held decides whether it is ever consulted.