The Wall That Leaked Only in Wind-Driven Rain

Why this matters

A leak that appears only in some storms gets a repair that is "confirmed" by the next few weeks of dry weather, which confirms nothing. In this case the first diagnosis was right: the leak came from a line-set penetration the shop had installed. The first repair, fresh sealant around the penetration, was the repair most techs would make. It failed anyway, because the sealant was never the part of the wall keeping water out. What fixed it was a staged water test that showed where the water actually entered, and opening the siding to find a flashing patch lapped backwards.

The signal

Two years after the shop installed a ductless system in a second-floor bedroom, the customer called about a brown stain on the drywall below the indoor unit. The line set, condensate line and interconnect cable passed through the west wall directly behind the unit, into a plastic line-set cover on the outside. The wall: fiber cement lap siding over a housewrap water-resistive barrier (WRB, the sheet behind the siding that sheds whatever water gets past it), over wood structural sheathing.

The customer's note was the most useful piece of evidence in the file, though nobody used it the first time: the stain grew only after storms that came in from the west or southwest. Ordinary rain, or a storm from the east, did nothing.

The first diagnosis, and the repair that was right about the place

The tech on the first callback went up a ladder to the penetration and found the sealant joint between the line-set cover's wall flange and the siding cracked along its top edge, with a gap about 1/16 in wide. The penetration was directly above the stain. The diagnosis was sound: the leak was at this penetration. The tech cut out the old sealant, cleaned the joint, and ran a new bead around the flange.

The next six weeks were dry, and the job was closed as fixed. Then a southwest storm came through with steady wind, and the stain was back, larger.

The location had been right. What was wrong was the model of how the wall keeps water out. Lap siding is not a sealed surface; in wind-driven rain, water is pushed through the laps and the butt joints and runs down the face of the WRB behind it. The sealant at the cover flange stops water at the face of the siding. It does nothing about water already behind the siding, and in a storm from the west, there is plenty of it.

The test that found the entry point

On the second callback, the tech ran a staged water test with a helper inside. Before starting, the system was switched off at its outdoor disconnect and breaker, because the interconnect cable through the same penetration carries line voltage on many ductless systems and water was about to be put on it, and the hose was kept off the outdoor disconnect and the exterior receptacle below.

The penetration sat about 13 ft up, so the spraying was done from the ground with a long wand rather than from the ladder, which kept the ladder's feet and rungs out of the water and kept the person holding a live hose off a wet ladder. The ladder went up only after the test, footed on dry ground beside the wet zone, for the inspection.

Method. A garden hose with a spray nozzle on a moderate fan, held about 2 ft from the wall, aimed upward and slightly into the laps to mimic wind-driven rain. Zones tested from the bottom up, so water from an upper zone could not run down and be blamed on a lower one. Each zone got 15 minutes. Inside, the helper read a moisture meter on its reference scale at the stain every 5 minutes, and watched with a flashlight.

Zone B is the one that matters. The joint the first repair sealed took 15 minutes of direct spray and the meter moved one point. Zone C, which put water on the siding above the penetration and let it run down behind, wetted the drywall in under 10 minutes.

A window sat about 4 ft above the penetration, a reasonable alternative source. After the wall dried, which took until the next day (the meter back to 12), the tech sprayed the window and its trim alone for 15 minutes. The meter held at 12. The window was ruled out, and the entry was between the window's sill and the penetration, behind the siding.

One more candidate was checked before opening the wall, because it is a common cause at line-set penetrations and it would have produced the same zone C result: a line set or sleeve that slopes down toward the inside carries water running down its outside surface straight into the room. A level laid on the sleeve read it sloping very slightly outward. Not the entry, though the repair still improved it.

What was behind the siding

The tech removed three courses of siding around the penetration by pulling nails rather than cutting, so no fiber cement was sawn. (Cutting fiber cement releases respirable crystalline silica; where a piece must be cut, use shears or score and snap, outdoors, rather than a dry saw, and where a saw is unavoidable use one with dust collection per the controls 29 CFR 1926.1153 sets for that task.) With the siding off, the problem was visible in a few seconds.

The original installer had cut an X in the housewrap, pushed the line set through, and folded the four flaps into the hole. Over that, a square patch of self-adhered flashing tape had been applied, and its laps were backwards:

  Section through the wall at the penetration

      WRB above
         |  water runs down WRB face
         v
   +-----+------+  patch TOP edge laid OVER
   |            |  the WRB above, held only by
   |  flashing  |  tape (now lifted 2 in)
   |   patch    |
   |    (O)  <--+-- line set through X-cut,
   |            |   flaps folded inward
   +-----+------+  patch BOTTOM edge tucked
         |         BEHIND the WRB below
         v
      WRB below  -> water on the patch face
                    runs behind this WRB

Correct lapping is upper layer over lower layer, everywhere, like shingles (the principle ASTM E2112 sets out for flashing openings): the patch's top edge should tuck up under the WRB above (through a slit), and its bottom edge should lie over the WRB below. Here both edges were reversed. Water running down the WRB from above met the patch's top edge, where the tape had lifted about 2 in, and went behind the patch to the X-cut, where the folded flaps funneled it straight into the hole. Water that stayed on the patch face reached its bottom edge and ran behind the WRB below, onto the sheathing. Two entries, both into the stud bay above the stain.

The sheathing below the penetration read high on the meter and was darkened in a fan shape for about 10 in below the hole, but was firm to a probe. No rot.

Why only some storms. In calm rain, very little water reaches the WRB behind lap siding, and what does runs past a reverse lap in amounts the wall can dry. In a storm with steady wind on the west wall, water driven through the laps keeps the WRB face running wet for hours, and a reverse lap delivers it inside the whole time. The customer's weather observation was a diagnosis in itself: the leak depended on the volume of water behind the siding, which means the defect was behind the siding.

The repair

The penetration was reflashed to shingle-lap order: the folded flaps trimmed back, a new flashing patch slipped up under a horizontal slit in the WRB above it and lapped over the WRB below it, a manufactured block for the penetration set into the patch, the line-set sleeve sloped slightly down toward the outside so water on it drains outward, and the WRB repaired and taped at its laps. The method belongs to How to Flash an Exterior Wall Penetration Into the WRB, which this repair followed rather than improvised. Sealant went on last, at the face, as the secondary line it is.

The siding went back with the original nails replaced by new ones of the type the siding manufacturer specifies, into the same studs. Before restoring the system, the tech confirmed the interconnect cable's jacket was undamaged where it passed through the new block, then closed the breaker and disconnect, ran the system, and checked that condensate drained from the line outside, which also confirmed the new slope had not trapped it.

How the fix was proved

The tech re-ran zone C, the one that had failed, for 15 minutes the next day after the stain area read 12 again. The meter held at 12 for all 15 minutes. Then zone B, for the same time: 12 throughout. Two weeks later, a west storm with steady wind passed through overnight. The customer reported no new stain, and a meter reading the next morning matched the reference spot elsewhere on the wall.

Only then did the shop repaint the stain, after the wall had read dry through one real storm, because paint over a damp spot hides the evidence of the next leak.

The lesson the shop took

The first repair was correct about where and wrong about what. The shop changed its callback note so that a wall leak at a penetration gets a water test before any repair, with the zones and the meter readings written down, and the job is not closed on dry weather. A repair that addresses the face of the siding is never the only repair at a penetration through lap siding, because the face is not where the wall's water control lives. See related: The Control Layers a Penetration Goes Through, which owns that model.

References

• Siding and WRB manufacturer installation instructions for penetration flashing and lap order
• ASTM E2112, Standard Practice for Installation of Exterior Windows, Doors and Skylights, whose shingle-lap principle for flashing is the basis most WRB manufacturers use at penetrations, where your code or manufacturer references it
• 29 CFR 1926.1153, respirable crystalline silica in construction, for cutting fiber cement siding
• See related: How to Flash an Exterior Wall Penetration Into the WRB; The Control Layers a Penetration Goes Through; How to Run a Line Set or Cable Through an Exterior Wall