Water Testing a Facade on Site

Studwork Starts Monday Whether or Not the Wall Is Proved

Metal stud starts on level four in nine days, and the drylining subcontractor has priced that floor assuming the wall he is building against keeps water out. Nobody has demonstrated that it does. The units went in five weeks ago and the only evidence anyone has is that there has been no serious weather since. The main contractor is not after reassurance. He wants a rig, a witness and a report, because from the moment a track is shot down against that wall the cost of finding a leak is multiplied by everything standing in front of it.

What he gets back is usually an argument, and it is almost never about whether the wall leaks. It is about which test was bought. A facade specification typically carries three or four testing clauses written at different times by different people: a mock-up clause in the performance section, a field testing clause in the execution section, a line in the quality assurance schedule about the installer's own checks, and a sentence in the preliminaries requiring weathertightness before follow-on trades. Those describe at least three different exercises, at different pressures, with different people watching and different meanings attached to the word fail. Working out which is being invoked is done at a table, not on a stage.

The other thing to be clear about before anyone books a bowser is what a field test can prove. It examines the installation, in the area you chose, on the day you chose it, at a pressure lower than the one the product was qualified at in a laboratory. It is a check on workmanship and interfaces — not a re-qualification of the system, not a warranty, and a clean result on eight bays out of four hundred is a sample result with all the limits that implies.

One of Them Is a Hose; the Rest Need a Fan

The methods available for a wall already standing on a building are a small family and they are not interchangeable. At one end sits a calibrated nozzle traversed along a joint by one person while another watches from inside — cheap, quick, portable, and applying no pressure difference across the wall at all. At the other sits a sealed chamber taped to the interior face, a spray rack outside delivering a calibrated rate of water, and a fan pulling the chamber down to a stated pressure. Most of the confusion on most jobs lies between those two, because both get written up in a site diary as a water test and only one produces a number a specification can be checked against.

A second family exists and it belongs to the mock-up, not to the building. Water penetration there is measured under ASTM E331 and, cyclically, ASTM E547; the companion methods for air and for structure are ASTM E283 and ASTM E330; AAMA 501 gathers them; and the European route runs through BS EN 12155 with BS EN 12154 supplying the classification a result gets reported against. Every one of those qualified the system before a bracket was cast in, and none of them is repeatable at level four with the tower crane running. Which is why the field pressure is deliberately lower: AAMA 502 and AAMA 503 both set the field figure as a fraction of the laboratory water pressure — two-thirds is the figure those documents carry — with a floor beneath which it does not drop. Read the edition the specification names; the numbers have moved between editions, and a specification will sometimes state its own and override both.

Water, Weather and a Sealant That Has Not Cured Yet

The commonest avoidable fail on a facade test is a joint that was never given time to cure. A weatherseal silicone reaches tack-free in hours and full cure in days to weeks depending on temperature, humidity and joint depth, and those figures are on the manufacturer's technical data sheet rather than in anyone's head. Test a perimeter bead at three days in cold, dry weather and the water will find it, the report will record a failure, and a joint that was going to be serviceable gets cut out at somebody's cost. Put the earliest valid test date in the programme, not the earliest date the access is free.

Bear in mind, too, that a wall which passes today can be undone tomorrow. Every trade that follows has a reason to cut, drill or seal something on it: a bracket for a blind track through a mullion, a lightning conductor bond, a signage fixing, a bead of mastic run along the outside of a sill by somebody tidying up and closing the weeps behind it. Test results have a date on them for a reason.

Weather governs the rest. A test near freezing is not a test — it is an ice hazard for everybody below the stage, and the methods set out the conditions they are valid in. Wind matters too: a gusting elevation makes chamber pressure unstable and the spray pattern uneven, and both push the result toward a fail nothing in the wall caused.

Then the logistics, which are why facade tests slip. A calibrated spray rack under ASTM E1105 delivers 3.4 litres per square metre per minute — five US gallons per square foot per hour, the same rate the laboratory methods use. A ten square metre test area is taking thirty-four litres a minute, so a fifteen minute run puts better than five hundred litres onto a facade with a building underneath it. That water needs a supply that holds the rate steady, a route down the elevation that does not end in a lift pit, and protection over anything finished below. The rack is outside on a stage while the observer is inside with the chamber: two crews, two radios, and an access plan written weeks earlier.

The inside has to be observable too. Back pans open, no drylining, no ceiling, working light, and the interior face and floor dry and photographed before anything starts. A stain that was already there is the beginning of a fortnight of correspondence.

Whose Signature Turns a Result Into Evidence

Three different exercises get called a water test on one job, and they are told apart by who is standing there. The first is the installer's own check, run by his foreman with a nozzle on a length of joint he is not sure about. It is legitimate, it is what AAMA 501.2 exists for, and it is the cheapest quality control on the job — but it is not evidence, and offering it as the proof a main contractor asked for is the fastest way to lose credibility. The second is a specification-mandated performance test: an independent agency, a named method, a named pressure, calibrated instruments, witnessed. The third is a forensic investigation after something has gone wrong, following AAMA 511 or ASTM E2128, which answers a different question altogether.

For the middle one, independence is written into the contract, and it means the agency is engaged by and reports to the owner or the main contractor rather than the facade installer. That is not an insult to the installer; it is what makes the report usable in a dispute two years later. The specification also names who pays — commonly the contract carries the scheduled tests while the responsible party carries everything after a failure, which is why the first conversation after a fail is about responsibility rather than water. Find that clause before the rig is booked.

Notice is the requirement most often broken. A witness who was not given the notice period the specification requires has not witnessed anything, however many people happened to be watching, and the test gets repeated. Calibration is the second: the nozzle's pressure gauge, the chamber's manometer and the flow measurement all have certificates, and the certificates belong in the report rather than in a drawer at the agency.

Be precise about what the witness is attesting to, because people sign these believing they mean more than they do. The signature says a stated area was tested by a stated method at a stated pressure for a stated duration, and that these observations were made. It does not say the wall is fit for purpose, it does not reach the four hundred bays nobody tested, and it does not transfer responsibility for the design.

One Fail Is a Sample Result, Not an Incident

A failure is not a defect in one bay. It is a defect found in an area chosen to stand for a population, and the specification will normally respond by expanding the sample: repair the failed area, retest it, and test additional areas at the client's selection. That is arithmetic rather than punishment — the sample was meant to represent four hundred units, one contained a defect, and the confidence it was supposed to buy has gone. Expect the extra bays to be chosen by whoever was least happy with the original selection.

What decides how bad the day is comes next: one-off or habit. A missing end dam in one transom is a defect. A gasket splice every glazier on the job has been forming the same way, a stack joint sleeve consistently set dry, a perimeter bead applied to a substrate nobody primed — those are systemic, and retesting will not fix them, because each test only ever finds the bay it is pointed at. A systemic finding is a re-work programme, and the honest answer to the question about releasing level four is no. Saying that on the day costs far less than saying it after the ceilings are in.

One field discipline gets broken almost every time and should not be: do not repair during a test. A bead run into a joint wet from your own spray rack will not adhere, and what it does instead is hide the leak convincingly enough to pass the retest and reappear in the second winter. Stop the water, let the joint dry to the manufacturer's stated condition, prepare and prime the substrate, repair, cure, then retest. The sequence takes days and the shortcut takes twenty minutes, which is why the shortcut keeps happening.

Finding which joint is at fault is a separate exercise from the test that failed, and it runs the other way. A performance test wets a whole area at once and tells you only that something in it leaks; a diagnostic run isolates. The ladder below works because water travels down and in.

1. Photograph the interior wet, and mark where the water presented and how far into the run — the entry point is almost never above where it showed.
2. Mask the whole test area with polythene and tape, sealed at the top, and confirm with a short run that a fully masked area stays dry.
3. Unmask the lowest joint only — the sill, the sub-sill or the transom gutter beneath the bay — and rerun for the method's stated period.
4. If it stays dry, remask and unmask the next element up: the vertical joints, the horizontal above, the perimeter seal at the jambs, then the head.
5. Change one element per run and write down every run, including the dry ones, because a dry run on a joint is the evidence that clears it.
6. Once water reappears, open the wall there rather than reasoning about it; a cover cap or a section of gasket will show an unsealed splice or a blocked weep in a minute.
7. Record the defect, then check the same detail in three bays nobody has tested. That check tells you whether you have a defect or a habit.

The Report Is the Thing You Are Buying

What survives all this is a document, and a document without the pressure on it is worthless. A usable field test report names the method and its edition, the pressure and how it derives from the specified laboratory figure, the spray rate and how it was verified, the duration and procedure, the exact area on a marked-up elevation, the weather and interior conditions, the calibration records, the observations with times against them, and everyone present with the party each was there for. A report saying a bay was tested and passed, with no pressure written on it, has recorded nothing checkable, and a pass at a pressure below what the specification required is a failure nobody has noticed yet.

File it against the mock-up report rather than on its own — the field figure was derived from the laboratory one and the two only make sense as a pair. Then hand the set into the operation and maintenance information with the remediation records: what leaked, what was found when the wall was opened, what was done, and the retest that closed it. A wall with that file behind it answers a leak investigation years later in an afternoon.