Drywall Partitions and Fire Compartmentation

It is the commonest and most underrated trade in contemporary building. A drywall partition is perceived as a light divider to run up quickly, whereas it is a layered system in which every component has a measured role: the framing gives the geometry, the wool dissipates sound energy inside the cavity, the boards give mass and closure, and resilient strips stop noise from bypassing everything through the structure. This is precisely the most misunderstood point: the declared performance belongs to the tested system, with those boards, that stud spacing, that type of screw and that infill. Replacing a component with one equivalent on paper does not preserve the result, because acoustic and fire behaviour arise from the interaction between layers, not from the sum of their qualities. A designer who specifies one system and a contractor who builds another are handing over a wall whose real performance nobody knows. Acoustics is almost always lost through flanking transmission rather than through the wall itself: sound travels via the continuous screed, via the suspended ceiling that runs over the partition, via the socket box facing the next room. This is why an acoustic partition is closed up to the structural slab and not stopped above the ceiling, and why boxes are never placed back to back in the same bay. Fire compartmentation adds a further logic: here it is not only the build-up that counts, but the penetrations. Every pipe, cable and duct crossing a fire wall opens a breach that must be closed with a system certified for that kind of penetration, and the documentation of those seals is as much part of the performance as the boards. The best wall in the world, pierced and unsealed, compartments nothing.

It is the discovery that infuriates, because it arrives when the wall is finished and the client still hears the neighbour. Sound is lazy only in appearance: it seeks the least resistant path, and if the wall is well built that path becomes everything else. The continuous screed running from room to room under the partition transmits beautifully; the suspended ceiling passing over the wall creates a duct in the plenum; the socket box facing the other room is a slot; and even a profile rigidly fixed to the slab carries noise through the structure without ever crossing the wool. This is why performance measured on site is nearly always lower than in the laboratory: in the laboratory there are no flanking paths. The countermeasures all belong to phase 01 and cost almost nothing if taken in time: cutting the screed under the wall, closing the partition up to the structural slab, staggering the boxes, resilient strips around the perimeter.

This is the source of half the disputes in this trade, and it is an honest misunderstanding: you read a performance figure, associate it with the board and assume a board of equal thickness and class gives the same result. It does not work that way. The declared value belongs to a precise configuration — that type and thickness of board, that number of layers, that stud spacing, that density and thickness of wool, that type and spacing of screw — because acoustic and fire behaviour arise from the interaction between layers: from the oscillating mass, from the air spring between the faces, from the way heat crosses and degrades each material over time. Change one element and the whole balance shifts, sometimes for the better and more often for the worse, but in any case unpredictably by hand. In practice: if the design specifies a system, build that system; if you want to change it, ask the manufacturer for a tested alternative configuration, and document it.