Equipment Pad Plumbing Leak Repair

Purpose

This procedure guarantees that a pad leak is identified by SIDE before anything is opened, repaired to a joint standard that survives the season, and released on a measurement rather than on a dry slab. Two failures make up almost every callback here. The first is a technician who tightens a wet union, watches the drip stop, and leaves a pool still losing water underground. The second is a fresh solvent-welded joint pressurized before the cement cured, which lets go days later with nobody on site and runs a pump dry until the seal cooks.

Scope

Covers locating and repairing a leak in the rigid PVC, unions, threaded adapters and valve bodies between the pool penetrations and the returns on a residential or light commercial pad, plus the pressure test that releases the work.

It does not cover the shell, the underground lines or the bucket-test method, which the leak detection SOP owns; the multiport valve or filter tank body, which the filter tank and multiport repair SOP owns; or the pump seal and heat exchanger, each covered elsewhere.

Roles and responsibilities

Which side of the pump the leak is on decides everything

The pad has two pressure regimes and they produce opposite symptoms, so the first reading is a side decision, not a repair decision.

Suction side runs from the skimmer and main drain to the pump inlet and sits BELOW atmospheric pressure while the pump runs. A hole there pulls air IN, it does not push water out: bubbles in the pot, air at the returns, prime lost after a shutdown, and a filter gauge BELOW its recorded clean baseline because the pump is moving less water. A pad normally sits above pool level, so a suction leak above the waterline loses no water with the pump off. It explains bubbles and cannot explain a dropping pool.

Pressure side runs from the pump discharge through the filter, heater, feeder and valves to the returns, above atmospheric while running. A hole there pushes water OUT while running and stops when the pump stops.

So: wet while running is pressure side, air in the pot with a dry slab is suction side, and water loss with a dry slab on both counts is not a pad leak at all. A technician who skips this and tightens the nearest fitting is repairing the joint the puddle ran toward, which on a sloped slab is regularly not the joint that leaked.

Two shop defaults make the rest decidable. Both are ours to set, not code figures. Pressure test: 10 psi of air on the isolated segment, held within 1 psi over 15 minutes. Loss gate: a net drop above 0.25 in per 24 hours, pool against bucket, is real loss. Tune both to your gauges and climate, and write down which you used.

Procedure

1. Open the pump disconnect, lock and tag it, and prove the motor terminals dead with a meter proved on a known live source immediately before and after. Accept when every conductor combination reads zero and the meter then re-reads correctly on the known source. Wrong is a dead reading from a dead meter, indistinguishable from a safe circuit until your hands are inside a cut pipe. A timer or controller "off" is not an isolating means, and a scheduled start with your arm in the suction line is what this step stops. Work practices are 29 CFR 1910.333(b)(2), not 1910.147, which excludes electrical utilization work at its own (a)(1)(ii)(C); the proving sequence is NFPA 70E-2021, 120.5.

2. Run the system once and record the side evidence before isolating anything. Accept when the ticket carries a wet-or-dry slab observation at 30 minutes of run time, the pot's air condition, and a filter gauge reading with its recorded clean baseline beside it. Wrong is deciding from where the water pooled. Hazard is a running motor and a pressurized tank: keep sleeves clear of the motor fan end, and do not crack the air relief or the pot lid while the system is under pressure.

3. Quantify the loss against the bucket baseline before repairing anything. Accept when you have a net figure (pool drop minus bucket drop) in inches per 24 hours plus its gallon equivalent: surface area in square feet times net inches divided by 12, times 7.48 gallons per cubic foot. Wrong is accepting the customer's estimate, which mixes evaporation, splash-out and a real leak in unknown proportions. Stop and defer the repair if no baseline exists; without a before number you cannot prove the after number. Hazard: none beyond a wet deck, so walk the perimeter rather than stepping across coping.

4. Shut down, relieve pressure, close the isolation valves and drain the segment before a union is cracked or a saw touches pipe. Accept when the air relief stops discharging, the gauge reads zero with the relief open, and the pot and drain plugs run dry. Wrong is breaking a union under head, which sprays at face height. If a heater fired within the hour, let the exchanger water cool first, because it scalds through a glove; where a feeder sits in the segment, treat the discharge as chemically loaded and wear sealed eye protection, not glasses.

5. Cap the isolated segment, pressurize to 10 psi of air, and locate the leak with soap solution. Accept when the gauge holds within 1 psi over 15 minutes, or when a joint bubbles and is marked. Wrong is including a filter tank, heater or valve body rated below the test pressure. Air in plastic pipe is stored energy and plastic fails by shattering, not weeping: keep everyone out of line with the end caps and the tank, hold the pressure low, and never leave a pressurized segment unattended. Stop rule: if it holds and nothing bubbles, the pad is not the leak - close the test, restore per step 7, and hand the ticket to the leak detection SOP.

6. Repair by joint type, and never past that joint type's limit. A weeping union gets a new o-ring in a clean groove, silicone lubricant only, nut hand tight plus a light turn; a wrench on a plastic nut is what cracked the last one. A threaded plastic connection gets PTFE tape or a plastic-rated sealant, one and a half to two turns past hand tight, never petroleum dope, which stress-cracks the fitting weeks later. A split fitting or pipe is cut out and replaced with couplings: cut square, deburr, dry-fit, then primer and cement per the manufacturer. Accept when the joint is square, fully bottomed and shows a continuous bead. Cement and primer are flammable and their vapor (typically tetrahydrofuran, methyl ethyl ketone, cyclohexanone) is an inhalation hazard: work with air moving and no ignition source, and inside an enclosed pump room ventilate first - if the SDS calls for a respirator, the work waits for a written program under 29 CFR 1910.134. Never torch or heat-gun PVC: it releases hydrogen chloride and never recovers its strength.

7. Cure, then restore water, then restore energy, in that order. Read the cure interval off the cement manufacturer's set-and-cure table at the pipe size, recorded ambient temperature and the pressure you will apply, and record which row you read. Accept when that interval has elapsed, the test caps are off, the valves closed in step 4 are reopened, the pot is filled and its lid o-ring silicone-lubricated, the bonding conductor at any disturbed component is landed on a clean lug, and the lock and tag come off last. Wrong is starting at full speed on a fresh joint. Start at the lowest available speed with the air relief open until solid water passes, and stand to the side of the tank and the repaired joint, because a joint or clamp that lets go throws the lid.

8. Verify with the same two measurements you opened with, not with a dry slab. Accept when the slab is dry at 30 minutes at full speed, the gauge reads within 1 psi of clean baseline, the pot is clear of bubbles, and a repeat bucket pair over the next 24 hours returns net loss at or under 0.25 in. Wrong is closing on the visual alone. Stop rule: pad tight and net loss still above 0.25 in per 24 hours means the ticket converts to a leak detection visit with the passing test attached, rather than closing complete. Hazard: the system is pressurized again, so shut down before cracking any relief or union to re-check.

The record this produces

One leak block on the work order, and a line on the pad's equipment record.

• Side evidence: slab wet or dry at 30 minutes running, pot air condition, filter psi with its clean baseline.
• Loss measurement: pool drop, bucket drop, net inches per 24 hours, surface area used, gallon equivalent - before AND after.
• Test result: segment tested, what was excluded and why, start pressure, pressure at 15 minutes, pass or fail.
• Repair: joint type, parts, cement and primer lot, the cure-table row read and the elapsed cure time.
• Restore: valves reopened, caps off, bonding conductor condition, lock removed, start speed, first-run observations.

The before-and-after loss pair is the only thing that answers the customer's real question months later, and it is the line shops skip. The cure-table row matters the day a joint fails at four weeks: a recorded row at a recorded ambient is a product question, a blank is a workmanship question.

One pad leak, filled in

A 16 by 32 ft rectangular pool, 512 sq ft of surface, about 20,000 gallons. Complaint: wet pad and a dropping level. Filter clean baseline on the equipment tag is 11 psi.

Step 2 settles the side. At 30 minutes of run time the slab is wetting under the heater outlet, and the same area is dry an hour after the pump stops. Filter reads 11 psi, on baseline, and the pot is clear. Pressure side.

Step 3, from the prior visit's bucket pair: pool dropped 0.95 in over 24 hours, bucket 0.20 in, so net is 0.75 in. In gallons, 512 times 0.75 divided by 12 is 32.0 cubic feet, times 7.48 is about 239 gallons in 24 hours. Far above the 0.25 in gate, so real.

Steps 4 and 5 run clean. Segment isolated from pump discharge to return manifold with the filter tank valved out, its body rated below the test pressure. Ten psi in, 9.6 psi at 15 minutes, a drop of 0.4 psi, inside the 1 psi allowance - and soap solution still bubbles steadily at the heater outlet union, whose plastic nut carries a hairline crack from over-tightening. A small leak can sit inside a hold tolerance, which is why the soap pass runs even on a segment that holds.

Step 6: union nut and o-ring replaced, hand tight plus a light turn, no wrench. Step 7: nothing was solvent welded, so no cure interval applies; caps off, valves reopened, pot filled, heater bonding conductor landed, lock removed, started at low speed with the relief open until solid water passed.

Step 8 fails. The slab is dry at 30 minutes at full speed and the gauge reads 11 psi, but the repeat bucket pair gives pool 0.85 in, bucket 0.20 in, net 0.65 in - 512 times 0.65 divided by 12 is 27.7 cubic feet, times 7.48, about 207 gallons. The union was worth roughly 30 gallons a day of the original 239, and 207 is somewhere else. The stop rule is taken as written: the pad is verified tight, the ticket does not close, and it converts to a leak detection visit carrying the pressure-test result. The customer is told that in those numbers, rather than "we fixed a leak."

References

• 29 CFR 1910.333(b)(2) for electrical work practices at the pump disconnect; 29 CFR 1910.147 excludes electrical utilization work at its own (a)(1)(ii)(C).
• NFPA 70E-2021, 120.5, for the live-dead-live proving sequence.
• 29 CFR 1910.134 for the written respiratory program any SDS-specified respirator use depends on.
• The solvent cement and primer SDS and set-and-cure table, read at pipe size, ambient temperature and applied pressure.
• See related: the leak detection visit SOP for the bucket-test method and off-pad leaks; the filter tank and multiport valve repair SOP; the pump and motor replacement SOP for bonding and isolation detail.