Fitting Cavity Barriers in a Ventilated Rainscreen

It has to breathe until the day it must not

A drained and back-ventilated wall depends on a path from the base termination to the head, and a solid barrier at every compartment floor cuts that path into storey-height compartments. That is a real loss: the cavity is what dries the back of the panel and carries away what the wind drove through the open joints, and a storey-height cavity with no inlet and no outlet of its own is a void, not a ventilated cavity. This is the whole reason open-state barriers exist. They hold a rated gap open in service and close it with an intumescent when the intumescent reaches its activation temperature, so the cavity behaves as one continuous path for the fifty years nothing happens and as a set of sealed compartments for the twenty minutes something does.

The number that matters is not the gap you can see over the top of the barrier. It is the rated free ventilation area the barrier system publishes for that product at that cavity depth, and it governs the whole run, because it is the smallest cross-section anywhere in the cavity. A base termination sized generously and a barrier line throttling the cavity to a fraction of it is a wall that is ventilated on the drawing and is not ventilated. Work it the other way round: establish what the barrier leaves open first, then check that the terminations and the cavity depth are at least as generous, rather than sizing the terminations and hoping the barriers agree.

Choosing a solid barrier instead is legitimate and it changes the design rather than simplifying it. Every storey then needs its own inlet at the bottom and its own outlet at the top, which means openings through the panel line at every floor rather than only at the base and the head, and each of those openings is another detail, another closure product and another count. It is worth deciding this at design stage on the arithmetic rather than at delivery on what the merchant had.

Count the crossings before you count the metres

A barrier line is quantified twice: once in linear metres, which is what it is bought as, and once in crossings, which is what it costs to install. The second number comes straight off the carrier grid. A run of vertical rails at fixed centres across a known elevation width gives a rail count, and on a system where the rails are continuous through the barrier, that count is the number of cut-and-seal operations on every horizontal line at that elevation. Multiply by the number of barrier lines and it is usually the largest single labour item in the package, and the one most often absent from the bill entirely.

The alternative is worth pricing rather than assuming. A carrier system built in storey-height rail lengths lets the barrier run past the rail ends unbroken, which removes every crossing on the line and buys a clean, photographable, uninterrupted barrier. It costs more rail pieces, more brackets, a splice or expansion detail at every storey, and a different panel setting-out. On a facade with four barrier lines and forty-eight rails per line, that trade is a hundred and ninety-two sealed crossings against a few dozen extra brackets, and the answer is not obvious in either direction until both are counted.

Bracket-only crossings sit between the two. Where the rail is set off the wall on brackets and the barrier passes behind the rail, only the bracket arm crosses the barrier: a smaller perimeter to seal, but usually a higher count, because brackets sit at vertical centres along each rail and more than one can land near a barrier line. Establish which of the three cases applies before the take-off, because the three produce different sealant volumes, different labour, and different photographs at handover.

Sealing a crossing is a perimeter problem, not an area one

Sealant round a rail is quantified the way sealant round a pipe is: the perimeter of the thing passing through, multiplied by the gap left around it, multiplied by the depth the tested detail asks to be filled. The perimeter is where estimates go wrong, because a rail is not a pipe and nobody instinctively adds up its sides. A 60 by 40 millimetre rail section presents 200 millimetres of perimeter at every crossing. Forty-eight of those on one line is 9.6 metres of sealed perimeter, four lines on the bay is 38.4 metres, and that is one bay of one elevation before any bracket, flashing or membrane fixing is counted.

The gap and the depth are not yours to choose. They come from the barrier system's tested detail, which will state the maximum annular space it was proven at and the fill depth that goes with it, and a crossing wider than that maximum is outside the evidence rather than merely untidy. The sealant itself has to be the product the barrier system was tested with or one its manufacturer accepts in writing; a tube of fire-rated mastic from the van is a different material with a different expansion behaviour, and the fact that it is red proves nothing. Where a crossing turns out oversized on site, the fix is a packing detail from the manufacturer, not more sealant.

Compressed, not cut to fit

A cavity barrier holds its position and its seal by compression. It is supplied wider than the cavity it goes into, squeezed in, and the residual pressure is what keeps it against the backup wall and against the back of the panel line for the life of the building. That is why a barrier cut neatly to the measured cavity depth is a failed barrier from the first day: it will sit there looking correct, and it will drop or gap as soon as the wall moves, the wool relaxes, or somebody leans on the scaffold beside it.

The complication is that a rainscreen cavity is not one width. The backup wall was built to a construction tolerance, the brackets were shimmed to bring the rail plane flat, and the insulation behind may be a nominal thickness that a compressed board does not hold everywhere. The barrier has to still be compressed at the widest point of the run and still be within its evidence at the narrowest, so the number to design against is the widest measured cavity on the line rather than the nominal depth on the section drawing. Measure it at both ends and the middle of every run before ordering; a run that varies by ten millimetres over eighteen metres is normal, and a barrier ordered for the average will be loose at one end.

Compression also changes what you buy. It is expressed as a fraction of the supplied width, not of the cavity, so the surplus is a division rather than an addition: a barrier compressed by a fifth needs material a quarter wider than the gap, and one compressed by a quarter needs material a third wider. On a few hundred metres of line that difference is a delivery, and it is the line a take-off written off the cavity depth alone will be short by every time.

The calculator below was written for a head-of-wall joint, and the mapping onto a cavity barrier is deliberate rather than accidental — it is the same geometry. The nominal width is the cavity depth the barrier fills; the run length is the barrier line; the backing depth is how far the barrier extends along the wall, which is the lamella height for an open-state product; the compression is the figure in the installation instruction. Read the movement class field as the tolerance band on your cavity rather than as a listed movement class, unless the system publishes one, and take the widest-width line out of the breakdown as the condition the barrier still has to be compressed at. One line needs watching as you do that: the uncompressed width it reports is worked from whatever went into the nominal field, so it is only compressed by the stated fraction at that width — where the run measures wider, the widest reading has to go into the nominal field and the width re-read before anything is ordered.

1. Measure the cavity at the barrier line itself, at both ends and the middle of the run, instead of taking the depth from the bracket schedule.
2. Take the widest reading as the width the barrier must still be compressed at, and check it against the maximum cavity the barrier's evidence covers.
3. Set the supplied width from the compression the installation instruction states, remembering that compression is a fraction of the supplied width and so the surplus is a division.
4. Mark the line from the compartment floor behind it, levelled across the whole bay, before a single length is offered up to the wall.
5. Fit the full lengths first and leave every crossing until the run is set, so each cut is made against the rail as it actually sits rather than where it was drawn.
6. Seal the crossings and the end butts last, in the order the tested detail gives, and photograph each bay before the next lift of panels covers it.

Compression is still not a fixing

Every open-state barrier system also specifies mechanical fixings, at stated centres along the run and usually at every end and every crossing as well, and they are not optional reinforcement of the compression fit. They are what stops a barrier migrating under its own weight and under the pressure differences a ventilated cavity develops in a gale, and their pattern was part of the tested arrangement. The first question they raise is what they go into: a barrier fixed into sheathing board alone has no more capacity than the board, and the fixing schedule normally wants structure, a proprietary bracket, or a masonry backup with an anchor qualified for it. The second is that every one of those fixings punctures the air and water barrier on the backup wall, so the sealing of the fixing is part of the detail rather than an afterthought for whoever is next on the wall.

Counting them is a two-number multiplication: how many barrier lengths are in the line, and how many fixings each length needs. Barriers arrive in fixed lengths, so that first number is the line length divided by the supplied length and rounded up, per line, per elevation, and it is worth doing per elevation because a bay with more openings has more short lengths and short lengths still carry end fixings. The calculator below has its fields named for cladding panels, because counting panel fixings is the job it was written for, but the arithmetic is a unit count multiplied by a per-unit fixing count and it transfers exactly: enter the number of barrier lengths where it asks for panels, and the fixings per length from the barrier's installation instruction where it asks for the pattern.

The drainage layer runs straight past it

Where the rainscreen is an open-jointed panel on a deep cavity, there is nothing between the barrier and the membrane. Where it is an adhered system — brick slips, thin stone, a mesh-and-render carrier — the cavity usually carries a drainage mat or a dimpled mesh, and that layer crosses every barrier line on the wall as a continuous sheet. It is the component most likely to compromise a barrier line, and the least likely to appear on the fire drawings, because it was specified by whoever owned the waterproofing rather than by whoever owned the compartmentation.

There are only two answers and both have consequences. Stop the mat above and below every barrier, and the barrier is clean but the mat now needs a discharge at each interruption so water reaching that shelf has somewhere to go, and the cutting waste rises because the elevation is being cut into storey-height pieces rather than run off the roll. Run it through, and a continuous polymer sheet passes a compartment line, which is a question about the material's reaction to fire and about whether the tested arrangement included it — a question for the fire strategy and, on a relevant building in England, for regulation 7 as well. Neither answer is available to the installer on the day; both have to be settled on a drawing.

The quantity is the elevation area that actually receives the mat, plus a waste allowance that is genuinely higher on an interrupted layout than on a continuous one. Take it per elevation rather than for the building, because it is the elevations with the most barrier lines and the most openings that generate the offcuts.