Leak Detection Visit Standard

Purpose

A leak call ends one of two ways. Either the shop hands back a measured loss rate attributed to a named component, with the test that isolated it, or it hands back an opinion. The opinion produces the second visit, the argument about whether the repair worked, and the customer who calls the next shop with the same complaint. This procedure guarantees every leak visit leaves the property with a number in inches per day, a number in gallons per day, and a component named by an isolation test, so the repair can be quoted, verified against that same number afterward, and defended in a builder or insurance conversation.

It also guarantees the tech does not open a pressurized line with their face over it, which is the injury this work actually produces.

Scope

Covers the diagnostic visit on a residential or small commercial in-ground pool or spa: complaint intake, evaporation separation, static and dynamic loss measurement, dye work on the shell and fittings, individual line isolation and pressure testing, and the written finding.

It does not cover the repair, which belongs to the plumbing and shell repair procedures, and it does not cover a vinyl liner pool below the liner, where the failure is a seam or a gasketed fitting rather than a buried line and the liner repair procedure owns it. Structural crack evaluation past a hairline is an engineer's call. The technique article on leak detection covers how each test is performed; this standard covers the order they run in, what each must return before the next starts, and what gets written down.

Roles and responsibilities

Procedure

1. Shut off the autofill and the pressure-side cleaner booster before any measurement, and record the time. An autofill valve makes loss unmeasurable, because the pool holds level while the meter runs. Accept when the autofill is physically valved off at the supply, not just switched at a controller, and the tech has seen the valve handle. Wrong looks like a customer saying it is off while a float valve in the skimmer throat is still open. If you cannot positively isolate the fill, stop: no measurement taken on this visit is usable, so rebook rather than record numbers you will have to disown later.

2. Interview for the fill history and put it in inches or fill events, not adjectives. Ask how far the level drops between fills, how often they add water, and whether the loss changed with the season, a storm, or an equipment change. Accept a statement you can convert to a rate: "down to the bottom of the tile in three days" is usable, "it seems low" is not. Wrong looks like a history built entirely inside the autofill era, in which case the water bill is your only history. Where loss started the week after a new pump went in, note it: higher return pressure opens marginal joints that held at the old flow.

3. Run the bucket test over a full 24 hours with the pump on its normal schedule. Set a straight-sided bucket on the second step, filled so the inside water level sits within about an inch of the pool level, and mark both levels with tape on the inside and outside. Accept the test when the bucket has not been splashed, the pool has had no swimmers, and no measurable rain fell. Net loss is pool drop minus bucket drop. Wrong looks like a bucket drop larger than the pool drop, which means splash-in or rain, so re-do rather than reason from it. A net loss under about a quarter inch in 24 hours is inside the noise of evaporation and wind for most climates: do not chase it, and say so plainly to the customer. Do not stand on wet coping to place the bucket; step down into the shallow end or use a pole.

4. Repeat the 24 hour test with the pump off the entire window. Same bucket, same marks, same conditions. Accept when both windows have valid bucket readings. The comparison is the whole diagnostic: markedly more loss pump-on points at the pressure side, meaning returns and any line downstream of the pump; loss that is the same or greater pump-off points at the suction side, the shell, or a fitting at the static waterline. Wrong looks like the customer restoring the schedule overnight because the water looked dull; if the pump ran at all, the window is void and repeats. Do not leave a pool with no circulation for more than about 24 hours in warm weather without telling the customer what that costs them in chlorine demand.

5. Dye test the shell, fittings, skimmer throats, and light niches with the pump off and the water still. Wait at least fifteen minutes after the pump stops so return flow has died, then release dye within an inch of the suspect feature. Accept a positive as dye drawing into a defined line or opening and continuing to feed, observed twice. Wrong looks like dye that fans and drifts, which is convection or residual current and is not a finding; record it negative rather than "maybe." Do not dive with weights to hold position at depth. A skimmer throat crack and a light conduit are the two positives most often missed, so test both even when the differential points elsewhere.

6. Isolate one line at a time and pressure test it, standing clear, with the pump locked out. Open the disconnect serving the pump and prove it dead before plugging anything, because a pump that starts against a plugged line drives pressure past the plug's rating in seconds. Plug at the wall fitting, connect the rig at the pad, and bring the line up slowly to a pressure within both the pipe's rating and the plug's rating, commonly 15 to 25 psi on schedule 40 PVC pool plumbing, never higher because the line "will not hold." Accept a line as sound when it holds the set pressure for 15 minutes with no more than a needle's width of drop. Wrong looks like a plug that creeps or a gauge falling to zero in under a minute, which is a seal failure, not a leak: re-do with the correct plug size and a wetted seal rather than raising pressure. Never stand over or in line with a plugged fitting while the line is pressurized, and bleed to zero at the gauge before removing anything.

7. Localize the failed line before you write the finding. A failed hold test tells you which run leaks, not where. Use a listening device on the deck over the line's path with the line held at test pressure. Accept a localization when the same point is loudest on two passes made from opposite directions. Wrong looks like a single loudest point found once and marked with paint; that is how a shop cuts a deck a body length away from the break. If two passes disagree, report the line failed and the location not established, and quote the repair with that stated.

8. Reconcile every finding against the measured loss before you leave. Add the losses your findings explain and compare them with the net loss from steps 3 and 4. Accept when the findings account for the measured loss within roughly a quarter of it. Wrong looks like a hairline dye positive offered as the explanation for several hundred gallons a day, which means a second and larger fault is still on the property. Say so on the worksheet and out loud: the shop that repairs the small one and closes the ticket owns the callback.

Converting inches to gallons, and why the customer needs both

Inches per day is what the customer can check themselves. Gallons per day is what makes the repair worth doing and what proves it worked. Convert with the pool's surface area: one inch of depth over one square foot is 1/12 of a cubic foot, and a cubic foot is 7.48 gallons, so gallons per day equals surface area in square feet times inches per day, divided by 12, times 7.48. State the surface area you used, since a pool with an attached spa sharing the water surface is not the pool's rectangle.

The record this produces

The leak worksheet, one per visit, attached to the job. Fields: surface area used and how it was derived; autofill isolation time and method; both windows with date, pump state, pool drop, bucket drop, and net inches; both losses converted to gallons per day; dye results by feature, each positive, negative, or not accessible; per-line pressure results with set pressure, hold time, and observed drop; localization method and whether two passes agreed; the reconciliation line; and the technician's stated confidence.

The service manager reads it to quote. The repair tech reads it to know where to cut. The office reads the net loss back to the customer after the repair, from the same worksheet, which is the only reason a post-repair bucket test means anything. An inconclusive test is a field entry, not an omission: a blank slot months later is indistinguishable from a test nobody ran.

Worked pass: a rectangle losing water only when the pump runs

Rectangular pool, 16 by 32 feet, 512 square feet of water surface, no attached spa. Customer reports adding water twice a week.

Step 1: autofill valved off at the hose bib, 8:10 a.m., handle photographed. Step 2: history converts to roughly three inches a week, which is usable.

Step 3, pump on its normal eight hour schedule: pool down 5/8 inch, bucket down 1/8 inch, net 1/2 inch in 24 hours. Converted: 512 times 0.5, divided by 12, times 7.48 equals about 160 gallons per day. That is above the quarter inch noise floor, so the visit proceeds.

Step 4, pump off the whole window: pool down 1/4 inch, bucket down 1/8 inch, net 1/8 inch. Converted: 512 times 0.125, divided by 12, times 7.48 equals about 40 gallons per day. The differential is 1/2 minus 1/8, so 3/8 inch per day, about 120 gallons per day that only leaves when the pump runs. Two findings are now expected, not one: something on the pressure side carrying 120, and something static carrying 40.

Step 5: dye at the shallow return draws in and holds, twice. Dye at the skimmer throat fans and drifts, recorded negative. Main drain not accessible without diving, recorded not accessible.

Step 6 FAILS ITS FIRST ATTEMPT. The return line is plugged at the wall and brought to 20 psi; the gauge falls to zero in about 40 seconds and the plug is visibly creeping. Under the step's own stop rule that is a seal failure, not a leak, so pressure is not raised. The plug is swapped for the correct size, the seal wetted, and the test re-run: the line now takes 20 psi and falls to 8 psi in about four minutes. Failed line established. The suction line and the second return each hold 20 psi for a full 15 minutes with no visible drop.

Step 7: listening from the pad end puts the loudest point about 11 feet along the run; listening from the pool end lands within a foot of it. Two passes agree, so the location is established and marked.

Step 8: the failed return explains the 120 gallons per day that appears only under pump pressure. It does not explain the 40 gallons per day measured static. The main drain was recorded not accessible, so the worksheet says the visit did not close that question and the quote carries a line for a return dive to dye it.

References

• ASTM F1346, in the edition your cover is labeled to, for the related safety cover work performed on the same property; not a leak standard, cited only where a cover is removed for access.
• 29 CFR 1910.333(b)(2), general industry practice for de-energizing before work on the pump circuit at the equipment pad; construction-scoped work falls under 29 CFR 1926.417.
• Pipe and fitting manufacturer documentation for the pressure rating of the plumbing being tested, and the test plug manufacturer's rated working pressure.
• See related: the PoolService leak detection technique article for how each individual test is performed, and the drain and acid wash standard for the case where the pool must be emptied.