The PEX Crimp That Let Go Behind the Vanity

Why this matters

Three weeks after a whole-house PEX repipe, a crimp joint behind a second-floor vanity separated overnight. The homeowner found water coming through the kitchen ceiling in the morning. The joint had held a 100 psi pressure test on install day. Nobody knew which joint had failed until the wall was open, and nobody could say whether it had ever been gauged until the ticket was read. The repair took three trades and most of a week; the check that would have prevented it takes about ten seconds per ring. This write-up works backwards from the failed fitting and the job ticket, because the gaps in the record turned out to be the finding.

The morning of the call

Water and electricity first. The homeowner had already closed the main shutoff, which stopped the flow. Water had run down the inside of the wall and through the kitchen ceiling around a recessed light. Before anyone touched the ceiling or the fixture, the tech opened that circuit's breaker at the panel, tagged it, and proved the fixture dead with a non-contact tester and then a two-pole tester proven on a known live outlet before and after. The kitchen floor was mopped before anyone stood under the sagging section on a ladder. The section of wet drywall was pierced with a screwdriver over a bucket to drain it rather than let it drop.

Then the source. With the vanity pulled, the wall behind it was wet from the floor to about 18 in up. A 12 in square cut at the hot supply stub showed a 1/2 in PEX hot line hanging free beside a brass crimp elbow. The pipe had pulled off the fitting with the copper ring still on the pipe.

The house had hot water off and the main closed. The tech restored cold water to the house by capping the hot riser in the wall with a gauged ring and a test cap, and opened the main. Every fixture's cold side worked; the cap was dry at 10 minutes.

What the fitting showed

The elbow, the ring and the pipe end were bagged and brought to the shop before anyone formed an opinion.

The ring. Viewed end-on it was visibly oval. With the go/no-go gauge, the GO slot slid over the ring in one position and would not slide over it 90 degrees away. One axis had been crimped to spec; the other had not. That is the signature of a crimp tool held off square to the pipe, which crimps an oval: tight across the jaws, loose across the gap.

The pipe. The inside of the pipe end carried barb marks deep on one side and faint or absent on the other, the side that matched the ring's loose axis. On that half, the pipe had been riding the barbs rather than being pressed down onto them.

The position. The ring sat about 1/8 in from the pipe end, inside the range the ring and fitting maker gives for its ASTM F1807 system, so placement was not the cause.

Why it held for three weeks. A crimp that grips on one axis can hold static test pressure for 15 minutes. What it cannot do is hold through thousands of cycles. This was the hot line: every draw heats the pipe and every stop lets it cool, and PEX expands and contracts with temperature far more than copper. Each valve closure adds a pressure spike. Over three weeks the pipe worked itself off the barbs on the loose side until it let go. That is consistent with the evidence; nobody watched it happen.

Reading a failed crimp tells you which question to ask next. Each way of getting a crimp wrong leaves its own marks, and the marks decide whether you have one bad joint or a day's worth:

• Under-crimped all the way round: GO fails in both positions, ring round. Suspect the tool's adjustment, and pull every joint that tool made since its last good check.
• Off-angle, as here: ring oval, GO passes one way and fails 90 degrees away, barb marks one-sided. A hand fault on that joint; the joints made just before and after it by the same tired hands are the suspects, not the whole job.
• Pipe not seated: a gap between the pipe end and the fitting shoulder, barb marks on the first barb only. Often comes in a run, where a tech pushed pipe on in a cramped bay.
• Ring off the barbs: ring too far back or hanging past the pipe end, with no barb under part of it. Placement, usually on fittings where the ring could slide before crimping.
• Over-crimped: NO-GO slides over, sometimes a cracked ring or a pipe cut by the ring's edge.
• Mixed systems: a copper ring on a fitting the pipe maker lists only for stainless clamps, or a fitting standard the pipe's print does not list. Every joint of that combination is suspect.

What the ticket showed

The repipe ticket was the shop's own, three days long. Read in order, it showed the tool was proven, the joints were made, and the test was passed. It also showed the hole.

  REPIPE TICKET, DAY 3 (excerpt, as filed)

  07:35  Crimp tool test on scrap: GO/NO-GO pass x2
  08:00  Joints 47-61 made   gauged 47-61 (15)
  12:30  Joints 62-71 made   gauged 62-71 (10)
  14:30  Joints 72-86 made   gauged: (blank)
  17:10  Pressure test: water 100 psi, 15 min
         start 100  end 100   PASS
  17:25  Day 3 total joints: 86   job complete

  Gauged on the job, days 1 to 3: 71 of 86

The morning check was done. The tool passed on scrap at 7:35 a.m., so the tool itself was not out of adjustment that day. The failed joint was a hand problem, not a tool problem, which matters for what the shop changed.

Joints 72 to 86 were never gauged. Fifteen joints, made in the last two and a half hours of a long day, with a blank where every earlier batch had a count. The vanity elbow was joint 79.

Nothing in the procedure caught the gap. The ticket recorded joint counts and gauge counts on separate lines, and nobody compared them. The pressure test passed, as the failed joint shows it could. The job was closed with 71 of 86 gauged, and 71 of 86 is written on the ticket in the tech's own hand.

The tech's account filled in what the ticket could not. The last batch went in as the homeowner was asking when the water would be back on. The gauge was in the tool bag in the basement. The tech intended to go back and gauge the batch before the test, and the test came first.

The other fourteen

One failed joint in an ungauged batch means fourteen more joints with the same history. The shop went back the following day.

• Eleven reachable without opening a finished surface, in the basement, the attic and under two sinks. Gauged in two positions: ten passed, one failed GO in one position. That one was a cold elbow in the attic, cut out and remade.
• Three behind closed walls, including the vanity's cold-side partner to joint 79. The homeowner agreed to three small access cuts, patched by the drywall contractor already coming for the main repair. Gauged: all three passed.

Ten plus one plus three is all fourteen. The batch finished at thirteen good joints and two failures, both from the same afternoon; the second had not yet leaked.

What the failure cost

The water. The pipe let go sometime after midnight and ran until about 6:30 a.m. at the hot line's full flow. The second-floor bath wall was wet to about 18 in, the vanity base had swollen past saving, and about 24 square feet of the kitchen ceiling below was stained and soft.

The drying. A water mitigation company set dehumidifiers and air movers for three days, opened the wet bath wall to 2 ft above the floor along 6 ft of wall, and logged moisture readings daily against a dry reference reading from unaffected framing in the same house, working to ANSI/IICRC S500, the water damage restoration standard that crew is trained to. Wet drywall and framing left closed for more than a day or two invite mold; the opened wall is the reason. The plumber's part of the drying was the moisture map at the first visit: a pin meter reading along the wall and the kitchen ceiling to show the mitigation crew where the water had traveled, so the wall was opened where it was wet and not just where the leak was.

The repair. The plumber's remake took under an hour: the elbow was replaced with a new ring gauged in two positions, the hot side refilled and purged at each fixture and checked dry at full pressure before the wall was closed. The kitchen light circuit stayed off and tagged until the ceiling was dry and replaced and the fixture was checked by an electrician where your state licenses that work. The drywall contractor replaced the kitchen ceiling section and the bath wall, the painter followed, and the homeowner bought a new vanity, which the shop covered. Across the plumber, the mitigation crew, the drywall contractor and the painter, the repair took roughly 30 labor hours over six days, against perhaps ten seconds to gauge the ring.

The relationship. The homeowner had paid for a repipe to stop leaks, and the first leak came from the repipe. The shop covered all of it without being asked to argue for it, which is the reason the homeowner left a fair review rather than a bad one.

The rule the shop wrote

The shop already required every crimp to be gauged; the procedure in How to Make PEX Crimp and Clamp Joints and Check Them says so. What failed was the distance between the rule and the moment: the gauge lived in a bag, and the check was batched to "later." Three changes close that distance.

• The gauge rides on the tool. Every crimp tool carries its go/no-go gauge on a short lanyard clipped to the handle. A gauge that is always in reach is used on the ring just made, not saved for a batch.
• Every ring is gauged in two positions as it is made, and counted. The ticket now has one line per batch: joints made, joints gauged, and the two numbers must match before the next batch starts.
• No pressure test until the counts match. The lead tech, or the tech on a solo job, reconciles the total joint count against the total gauge count before filling the system. A mismatch is a stop: gauge the missing joints, then test. On this job that rule would have stopped the 5:10 p.m. test at 71 of 86.

What it did not change. The tool was not the problem, so the morning scrap check stays as it was. The pressure test stays at 100 psi for 15 minutes as the shop default; it was never going to catch a joint that grips on one axis, and pretending it would is how the shop got here.

References

• ASTM F1807, metal insert fittings and copper crimp rings for PEX, and the pipe maker's installation instructions, which are part of the listing.
• ANSI/IICRC S500, the water damage restoration standard the mitigation contractor works to.
• See related: How to Make PEX Crimp and Clamp Joints and Check Them; Reading PEX Pipe and Fitting Markings Before You Join; The Joining Method Standard SOP.