Plumbing
Water-meter isolation tests for hidden leaks: a utility-based field guide
Trace unexplained metered flow through verified valve boundaries, with timestamped readings, meter limits and a record that separates refill episodes from continuous loss.
Direct answer
Start a water-meter isolation test with every water use stopped, photograph the reading, and compare it after 30 minutes, the interval Seattle Public Utilities uses for its leak check. If the meter advances, verify that the building shutoff actually closes, then compare readings with that valve closed to separate downstream building use from flow in the remaining metered branches. Use the meter's displayed volume increment and low-flow specifications to choose a longer observation window when needed; EPA WaterSense recommends a two-hour no-use check for homes.
On this page
What does a water-meter isolation test tell you?How do you read the meter leak indicator and register?What meter resolution and low-flow limits control the test?How long should the test run, and how do you calculate leak rate?How do you establish a true no-use baseline?What fixture and valve isolation sequence narrows the leak?Is the leak continuous or intermittent?Worked example: what do timestamped isolation readings show?What uncertainty requires utility testing or leak locating?How do you verify repair and restore service?How much water and money does the leak waste?What belongs on the field checklist and utility record?How should facilities and public plumbing bids scope the work?Open the related tool for the calculation, checklist, or packet.
FieldOS by AnvilfieldOffline-first FieldOS for job photos, proof packets, property memory, calculations, teams, billing, and owner-ready reports.What does a water-meter isolation test tell you?
A water-meter isolation test narrows active unexplained flow to a segment or branch before opening walls or ground. EPA's November 2023 WaterSense at Work guide treats a changed reading during a no-use period as evidence of possible leakage. Account for fixtures, automatic equipment and essential loads before assigning leakage. [S4]
Draw the measured boundary first: street main, meter, every tee before the building shutoff, building shutoff, then downstream fixtures. Mark irrigation, outbuildings, storage tanks, bypasses, separate services and fire supplies. An outdoor meter followed by an irrigation tee and then a building main leaves irrigation connected when that main closes; an indoor basement meter does not measure buried leakage before the meter.
Use the domestic service and meter guide for utility coordination. Responsibility follows the address's utility rules: Portland City Code 21.30.130 assigns domestic and irrigation service maintenance through the meter stop for connections of 1 inch or less to the Water Bureau; larger connections and vault installations have different termination points. Fire services use the property side of the curb valve. These are maintenance responsibilities, not a universal ownership rule. [S13]
Meter testing measures volume on a pressurized supply. A separate hydrostatic pressure test measures pressure behavior in an isolated section; pressure drop alone does not supply gallons per minute. Use low-pressure diagnosis for pressure symptoms and sewer camera inspection for drain-side defects.
How do you read the meter leak indicator and register?
Read the unit printed on the register, the complete cumulative volume and its smallest readable increment. Photograph the meter serial and decimal positions. A billing portal labeled CCF does not establish that the local register displays CCF.
Distinguish the cumulative totalizer, mechanical low-flow indicator, instantaneous flow screen and stored leak alarm. Portland Water Bureau describes triangular and snowflake indicators whose movement can be slow at small flows. A moving indicator shows water passage; an alarm summarizes a manufacturer's detection rule and history. [S3]
Anvilfield's identity check matches the meter serial to the account and tests a known downstream faucet. Confirm that this meter responds, then close the faucet and exclude that draw from the no-use interval. A blank or nonresponsive register needs utility investigation before a stationary reading receives a leak interpretation.
Seattle's 2026 rate schedule uses 1 CCF = 100 cubic feet = 748 gallons. Portland Water District in Maine supplies a useful billing example: 900 minus 897 HCF equals 3 HCF, or 2,244 gallons. [S9] [S8]
Neptune's February 2022 ProCoder)R900i sheet specifies eight-digit readings, 15-minute reading/flag intervals and 96 days of hourly history. Badger's November 2026 G2 manual shows that endpoint output can omit digits visible locally. Record local and remote resolution separately. [S6] [S10]
Record fresh readings of the same totalizer, separating sample and upload times. Resolve rollover capacity or meter-replacement readings before subtracting totals. Neptune's September 2021 FAQ also distinguishes the sweep hand from the wheel bank; never assign a gallon value to a triangle revolution without that register's documentation. [S5]
Badger's November 2026 G2 manual defines total consumption as forward volume minus reverse volume, so zero net change can conceal forward passage. Record direction reversals and current flow as well as the total. Investigate pressure behavior and protection through closed-system thermal expansion and backflow assembly testing; keep required backflow protection in service under Portland Code 21.30.320. [S13] [S10]
What meter resolution and low-flow limits control the test?
Three limits govern the result: displayed volume increment, meter-body flow performance and reporting interval. The increment controls displayed volume precision; body performance describes measurement during flow; reporting interval controls data availability. Identify the body, register, size, unit, endpoint and applicable firmware before applying another meter's numbers.
Neptune's July 2023 small T-10 sheet specifies 95% registration at 1/8 gpm for its 5/8-inch body, even though the paired ProCoder)R900i displays tenths of a gallon. A 0.1-gpm steady flow is below that published test point; waiting longer accumulates volume but does not change the body's low-flow performance. [S15] [S6]
| Meter and register | Size / configuration | Totalizer increment | Published flow performance relevant to low-flow work | Authority / edition |
|---|---|---|---|---|
| Neptune T-10 with ProCoder)R900i | 5/8 inch | 0.1 US gal; 0.01 ft³; 0.001 m³ | Body: 1/8 gpm at 95% registration; normal range 0.5 to 20 gpm at 100% ±1.5% | [S6], 02.22, p. 2; [S15], PS T10 SMALL 07.23, p. 4 |
| Neptune T-10 with ProCoder)R900i | 3/4 inch | Same unit-specific increments as above | Body: 1/4 gpm at 95% registration; normal range 0.75 to 30 gpm at 100% ±1.5% | [S6]; [S15], p. 4 |
| Neptune T-10 with ProCoder)R900i | 1 inch | Same unit-specific increments as above | Body: 3/8 gpm at 95% registration; normal range 1 to 50 gpm at 100% ±1.5% | [S6]; [S15], p. 4 |
| Neptune T-10 body | 1-1/2 inch | ProCoder)R900i: 1 US gal; 0.1 ft³; 0.01 m³ [S6] | 0.75 gpm at 95% registration; normal range 2 to 100 gpm at 100% ±1.5% | [S7], PS T-10 INTERMEDIATE 07.23, p. 3 |
| Neptune T-10 body | 2 inch | ProCoder)R900i: 1 US gal; 0.1 ft³; 0.01 m³ [S6] | 1 gpm at 95% registration; normal range 2.5 to 160 gpm at 100% ±1.5% | [S7], 07.23, p. 3 |
| Badger E-Series G2, bronze | 5/8, 5/8 × 3/4 or 3/4 inch | 0.01 US gal; 0.001 ft³; 0.0001 m³ | Minimum test flow 0.04 gpm; ±3% at that test point; ±1.5% over normal test limits: 0.08 to 30 gpm for 5/8 and 5/8 × 3/4; 0.10 to 35 gpm for ordinary 3/4-inch domestic models | [S10], ESM-UM-04284-EN-13, November 2026, pp. 15, 24 |
| Badger E-Series G2, bronze | 1 inch | Same totalizer increments as row above | Minimum test flow 4.5 gal/hour; ±3% at minimum test flow; normal test limits 0.16 to 62 gpm at ±1.5% | [S10], November 2026, pp. 15, 24 |
| Badger E-Series, cold-water engineered polymer | 5/8, 5/8 × 3/4 or 3/4 inch | 0.01 US gal; 0.001 ft³; 0.0001 m³ | Minimum test flow 0.05 gpm; normal test limits 0.1 to 25 gpm for 5/8 and 5/8 × 3/4; 0.1 to 32 gpm for 3/4 | [S11], ESM-DS-00041-EN-09, November 2022, p. 2 |
| Badger E-Series, cold-water engineered polymer | 1 inch | 0.01 US gal; 0.001 ft³; 0.0001 m³ | Minimum test flow 0.25 gpm; normal test limits 0.4 to 55 gpm | [S11], November 2022, p. 2 |
Badger's cited G2 bronze and engineered-polymer ratings apply to new meters at water temperatures of 45 to 122 °F. The G2 rows describe ordinary domestic configurations; separate residential fire-service configurations have their own ranges. The November 2022 polymer sheet specifies ±1.5% over normal limits and ±3% over its extended low-flow range. [S10] [S11]
Minimum test flow is a specified test point, not a guaranteed starting-flow threshold. Keep Neptune's percent registration distinct from Badger's percent error. In-service condition and actual flow during an episode govern the field interpretation: a low interval average can consist of higher-rate refills separated by zero flow.
How long should the test run, and how do you calculate leak rate?
Plan a 30-minute Seattle utility screen, a one-hour EPA commercial comparison or a two-hour EPA home comparison according to the setting. These are screening procedures, not competing legal minimums. Extend observation to resolve the display increment and relevant equipment cycles. [S1] [S4] [S2]
Calculate rate by subtracting the opening volume from the closing volume and dividing by elapsed minutes. Multiply a constant gallons-per-minute rate by the minutes in a day for its daily projection. Convert cubic feet before dividing; include the date when timing across midnight.
Method: Q = (V_end - V_start) / t; daily gallons = Q × 1,440. Here Q is gallons per minute, V is cumulative gallons and t is elapsed minutes. For a cubic-foot register, multiply the volume difference by 7.48 first, using Seattle's billing conversion. [S9]
Anvilfield's planning rule is to accumulate at least ten displayed increments for a quantitative rate estimate. For increment r and identical display conventions at both endpoints, the difference between rounded or truncated readings has quantization uncertainty below one increment; rate uncertainty is below r/t. Ten increments limit that effect to about 10% of the measured change, separately from calibration, low-flow error, timing and operating conditions.
| Display increment, US gal | One increment over 30 min, gal/hour | Ten increments, gal | Time to register ten increments at 0.1 gpm | Meaning of zero displayed change in 30 min, fresh readings with forward-only flow |
|---|---|---|---|---|
| 0.01 | 0.02 | 0.1 | 1 min | Less than 0.01 gal registered in that interval |
| 0.1 | 0.2 | 1 | 10 min | Less than 0.1 gal registered |
| 1 | 2 | 10 | 100 min | Less than 1 gal registered |
| 10 | 20 | 100 | 1,000 min, or 16 hours 40 min | Less than 10 gal registered |
The time column assumes the meter registers the planning flow. The 10-gallon increment belongs to Badger's 12-inch G2, whose minimum test flow is 8 gpm; a longer wait cannot establish accurate measurement at 0.1 gpm on that model. Divide ten increments by the planning flow to obtain the required time. [S6] [S10]
An unchanged display does not bound physical loss below the meter's measuring range. Report tiny changes in gal/hour when rounding to two decimal places in gpm would conceal them. When the required window disrupts occupancy, obtain a finer local indication or utility diagnostic data rather than imposing the table's longest shutdown.
How do you establish a true no-use baseline?
Establish the baseline at normal operating pressure after known startup demands have finished. Following a shutdown, let tanks refill and the supply stabilize before timing. Badger's G2 operating instructions distinguish a full operating meter from empty-pipe conditions; trapped air, incomplete filling and storage makeup defeat a simple steady-loss comparison. [S10]
Notify occupants of the exact start and finish times. Inventory kitchens, laundry, toilets, ice makers, reverse-osmosis reject water, humidifiers, pool autofill, irrigation, treatment regeneration and facility makeup or overflow loads. EPA's November 2023 commercial procedure uses off-peak testing with water-using equipment and sanitary use stopped. [S4]
Record disabled schedules and restore them afterward. Review softener and filtration operating cycles before interpreting an automatic discharge as leakage. Measure unavoidable essential demand separately or label it unresolved; the total meter change includes that demand until it is accounted for.
EPA WaterSense recommends checking for silent toilet leaks annually with a 10-minute tank dye test. [S2]
Use accessible domestic controls within the agreed boundary. A domestic main can also feed residential sprinklers: identify that connection before closure and arrange any impairment with the owner and fire-system contractor under local fire-department procedures. Seattle Fire Department's current Systems Testing guidance addresses notification, mitigation and restoration; essential process loads also need an agreed operating plan. [S18]
What fixture and valve isolation sequence narrows the leak?
Establish active unexplained flow, prove the proposed shutoff, then compare branch groups and individual fixtures. Start at the building main to separate indoor demand from upstream tees, or at fixtures to resolve a known refill problem. The valve selection and isolation guide covers the hardware behind those boundaries.
Portland Water Bureau's current procedure treats continued faucet flow after several seconds with the main closed as failed isolation. Use a cold-water faucet and distinguish brief stored-pressure or elevated-pipe discharge from sustained supply. Reclose it before timing; sustained supply prevents an upstream/downstream conclusion. [S3]
Anvilfield adds a slow-passage recheck when closure is doubtful: after pressure dissipates, repeat the controlled cold-faucet check. Pressure recovery alone also needs a thermal-expansion and stored-water check. Resolve passing valves or parallel feeds before assigning buried service work.
| Step | Action and recorded condition | Result | Interpretation | Next action | Basis |
|---|---|---|---|---|---|
| 1 | Stop or account for every water user; record the no-use interval | Register advances | Registered use needs explanation | Confirm unit, active flow and schedules | [S1], [S3], [S4] |
| 2 | Close building main; open cold faucet briefly to test closure; reclose faucet | Flow persists after stored discharge | Isolation unproven | Repair/verify valve or alternate supply before assigning location | [S3]; branch interpretation by Anvilfield |
| 3 | With closure proven, repeat timed meter check | Meter stops | Downstream contribution is supported if reopening restores flow | Reopen, confirm recurrence, then isolate fixture groups | [S1], qualified by [S3] |
| 4 | With closure proven, repeat timed meter check | Meter advances | Meter-to-valve piping or another still-fed branch carries flow | Identify pre-valve irrigation, outbuildings and bypasses; test those separately | [S1] plus topology rule |
| 5 | Shut toilet fixture stops, documenting each one | Rate decreases or stops | Toilet group contributes flow | Dye-test tanks individually for 10 min; inspect fill/overflow operation | [S2]; rate comparison by Anvilfield |
| 6 | Close the irrigation branch's water isolation valve, not only its timer | Rate decreases or stops | Irrigation branch contributes flow | Separate supply piping, valves and zones using branch access | Anvilfield boundary method; [S3] |
| 7 | Isolate nonessential appliance/treatment feeds with proper valves and controls | Flow changes | That feed contributes flow | Compare cycle logs, discharge and physical condition | [S4]; Anvilfield sequence |
| 8 | Investigate hot-water distribution under the installed equipment's shutdown procedure | Meter contribution changes | Hot branch requires further isolation | Follow the installed heater shutdown, relief-protection and refill instructions | [S16]; heater service |
| 9 | Reopen the suspected branch, repeat comparable interval, then restore normal valves | Flow returns with branch and falls when re-isolated | Repeatable branch association | Locate the fault and verify repair | Anvilfield repeatability rule |
Change one boundary between comparable intervals, log intentional draws outside them and record pressure where available. Pressure changes and interrupted refills alter rates. Repeat the suspected state after stabilization.
Closing a toilet stop prevents refill while tank-to-bowl leakage can continue from stored water. Start the dye check with a filled tank, add dye without flushing, inspect the bowl after EPA's interval, then flush to avoid staining. Multiple buildings on a master meter need branch shutoffs or submeters before consumption is assigned to one tenant. [S2]
A. O. Smith's November 2023 Series 120-121 condensing-gas manual requires a full tank before operation and an open cold supply during use; it prohibits plugging the relief valve or obstructing its discharge. Follow the installed heater's shutdown instructions before isolating its feed and its refill/restart instructions afterward. Continued hot-side supply after closure requires crossover or alternate-feed checks. [S16]
Is the leak continuous or intermittent?
Two meter readings establish accumulated volume and average flow, not whether flow continued throughout the interval. Continuous registered flow persists through the observed no-use window; intermittent flow occurs in episodes separated by intervals without registered flow. Record intermediate readings or current-flow observations to distinguish them.
Neptune's September 2021 ProCoder FAQ classifies usage in all 96 daily 15-minute intervals as a continuous leak flag and usage in 50 to 95 intervals as intermittent. Badger's November 2026 G2 manual defines suspected leakage as flow above its low-flow operating rate for 24 hours. Neither flag establishes a defect: intended use can trigger it, and fewer episodes can still waste water without triggering it. [S5] [S10]
A facility's cooling tower uses legitimate makeup water. Compare controller or building automation records with branch measurements before treating every occupied hour's consumption as a fault. Use in every reporting bin does not establish uninterrupted flow within those bins. [S4]
Anvilfield recommends an initial 24-hour schedule review, extended through the next documented regeneration or other less-frequent cycle. Record the actual sampling interval, upload delay and time zone. Request the finest available utility data.
| Date / local time | Window / data source | Boundary / equipment state | Volume or evidence | Pattern interpretation | Next check |
|---|---|---|---|---|---|
| 2026-11-05 08:00 to 08:30 | 30 min; direct register | Routine users stopped, fixtures connected | 12 gal registered | Unexplained volume; continuity unresolved by endpoints | Start Table E sequence |
| 2026-11-05 10:20 to 10:50 | 30 min; direct register | Building open; toilets and irrigation isolated | 0 gal change at 0.01-gal resolution | No registered change in tested boundary | Reintroduce one branch |
| 2026-11-05 10:55 to 11:25 | 30 min; direct register | Toilet group reopened; irrigation isolated | 3 gal registered | Toilet group contribution averages 0.1 gpm | Dye-test individual tanks |
| 2026-11-05 11:05 to 11:07 | 2 min; intermediate direct readings | Toilet group reopened; irrigation isolated | 123491.00 to 123494.00 gal: 3 gal in 2 min | Episode average 1.5 gpm | Confirm no flush; check dye and capture recurrence |
| 2026-11-05 10:55 to 11:05; 11:07 to 11:25 | 10 + 18 min; direct readings and current-flow observation | Same valve states | 123491.00 unchanged before refill; 123494.00 unchanged afterward; zero current flow observed | Isolated refill episode; full-window average 0.1 gpm | Dye-test and capture the next refill cycle |
| Next documented irrigation start | Controller record plus meter before/after | Irrigation reconnected for its authorized operating window | Record actual start/finish readings and zone | Scheduled demand versus unexpected between-cycle flow | Test branch with timer off, then with supply isolated |
| Next documented softener regeneration | Equipment event record plus meter | Other controllable branches unchanged | Record actual cycle times and meter change | Scheduled cycle or unscheduled discharge | Compare installed valve/controller instructions |
The worksheet uses Table E's inputs in America/New_York, EST (UTC-05:00). Reconnection filling finishes before 10:55; the later refill runs from 11:05 to 11:07, between two zero-flow periods. Exclude a flush, dye-test and capture recurrence before assigning a recurring toilet leak. Enter actual readings for irrigation and regeneration cycles.
Worked example: what do timestamped isolation readings show?
Anvilfield estimates a baseline average of 0.4 gpm from 12 registered gallons over 30 minutes, or 576 gallons per day if that average holds across 24 hours. The worksheet uses a 5/8-inch Badger E-Series G2 bronze meter, a local US-gallon display with 0.01-gallon increments, 60 °F water, forward-only flow and normal operating pressure. Its specified new-meter normal range covers the example's active flows. [S10]
The outdoor meter feeds an irrigation tee before the building main; the toilet group is downstream of that main. The cold-faucet closure check succeeds before the closed-main interval; irrigation scheduling stays off throughout. Seven 30-minute comparisons use Seattle's screening duration, with one five-minute transition between each pair. [S1]
Five transitions each contain 2 registered gallons from intentional draws, replenishment and ongoing flow. The 10:50 to 10:55 transition contains 3 gallons, including the toilet group's initial refill, completed before timing. Every timestamp below is November 5, 2026, America/New_York, EST (UTC-05:00). The cumulative readings and operating states are inputs to the Anvilfield estimates below.
| Date, local time | State during interval | Start, US gal | End, US gal | Elapsed | Change | Average flow | Decision |
|---|---|---|---|---|---|---|---|
| 2026-11-05 08:00 to 08:30 | Routine uses stopped; building, toilets and irrigation connected | 123450.00 | 123462.00 | 30 min | 12 gal | 0.4 gpm | Establish baseline average |
| 08:35 to 09:05 | Toilet group isolated; building and irrigation connected | 123464.00 | 123473.00 | 30 min | 9 gal | 0.3 gpm | Toilet-group contribution warrants individual testing |
| 09:10 to 09:40 | Building main proven closed; irrigation connected upstream | 123475.00 | 123484.00 | 30 min | 9 gal | 0.3 gpm | Flow remains outside closed building boundary |
| 09:45 to 10:15 | Building main closed; irrigation supply also isolated | 123486.00 | 123486.00 | 30 min | 0 gal | 0 registered gpm | Irrigation isolation removes remaining registered use |
| 10:20 to 10:50 | Building reopened; toilets and irrigation still isolated | 123488.00 | 123488.00 | 30 min | 0 gal | 0 registered gpm | No registered use with those branches isolated |
| 10:55 to 11:25 | Toilets reconnected and initial filling complete; irrigation isolated | 123491.00 | 123494.00 | 30 min | 3 gal | 0.1 gpm | Toilet-group volume recurs after initial filling; see Table D |
| 11:30 to 12:00 | All tested branches restored to baseline state | 123496.00 | 123508.00 | 30 min | 12 gal | 0.4 gpm | Original average repeats |
Divide each interval's volume change by 30 minutes. The seven timed windows total 210 minutes and 45 gallons; six five-minute transitions add 30 minutes and 13 gallons. Together they reconcile the four-hour schedule with the register increase from 123450.00 to 123508.00 gallons: 58 gallons.
The repeated states associate an average 0.1 gpm with the toilet group and 0.3 gpm with irrigation under comparable conditions. Closing the building main leaves the irrigation tee live, so continued flow at 09:10 does not establish a buried service fault. Closing irrigation removes the remaining registered use; reopening branches restores their contributions.
Table D separates the later toilet refill from reconnection filling. Confirm the tank defect with dye and another unprompted refill; isolate irrigation supply piping, valves and zones to locate its unwanted flow. Use observed daily event counts for loss projections and supply-side locating for the excavation point.
Resolution method: the 0.01-gallon increment divided by 30 minutes contributes less than 0.02 gal/hour of quantization per interval; subtracting two interval rates gives a conservative bound below 0.04 gal/hour. Retain raw readings; body accuracy, timing and operating changes remain separate.
What uncertainty requires utility testing or leak locating?
Resolve the isolation boundary before ordering excavation. A coarse unchanged register calls for better volume evidence; sustained flow through a closed shutoff calls for valve or topology work. A wet patch with no corresponding supply evidence calls for comparison with rain, drainage and groundwater conditions as well as supply-side investigation.
Report the supported segment: flow remains between meter and proven building shutoff after upstream branch isolation. Arrange meter testing and sealed-equipment work through the utility. Customer shutoff authority is local: Seattle permits customer operation of top shut-off meters but reserves older meter valves for SPU crews. Use the building valve for the ordinary test. [S19] [S3] [S10]
| Observation | Supported conclusion | Unresolved issue | Next action / responsible party | Basis |
|---|---|---|---|---|
| Steady advancing totalizer; every intentional use accounted for | Unexplained registered flow is active | Fault location | Compare branch isolation states; plumbing contractor | [S1], [S4]; Anvilfield interpretation |
| Building main closed; faucet still has sustained supply | Building isolation unproven | Passing valve or alternate feed | Verify supply topology and repair/test shutoff; plumbing contractor | [S3] |
| Proven building main closed; meter still advances; irrigation tee still connected | Flow remains in live metered section or irrigation | Which branch carries flow | Isolate upstream tee before condemning service pipe | [S1], [S3]; topology analysis |
| Identified branches isolated; meter advances | Remaining metered segment needs investigation | Valve leakage, overlooked branch or meter issue | Prove closure; request utility meter evaluation; then locate customer-side fault | [S3], [S10] |
| No change on a functioning 1-gal register in 30 min; current readings, forward-only flow | Less than one gallon registered in window | Small flow, low-flow under-registration, missed episode | Use finer indication or longer observation; check model limits | [S6]; Table B calculation |
| Decreasing total or direction reversals | Net registration differs from forward passage | Reverse flow, thermal expansion, installation or cross-connection issue | Utility and plumber check direction, pressure and required backflow protection | [S5]; [S10], pp. 15, 28; [S13] |
| Register unresponsive to known draw, or portal stale | Reading cannot establish current consumption | Wrong meter, display/endpoint fault or missing data | Match serial and local response; request utility investigation | Anvilfield identity check; [S10] |
| Wet ground or supply noise upstream of an indoor meter; no meter change | That pipe is outside the measured boundary | Supply leak versus drainage, groundwater or another source | Review ownership; locate supply-side fault; compare weather and drainage | Anvilfield topology rule; [S3] |
| Zero direct reading now; historical alarm remains | Current observation and stored alarm differ | Alarm clearing interval or recurring event | Read current flow; check OEM alarm definition and schedule | [S5], [S10] |
| Usage exists in every hourly bin at an operating facility | Recurring hourly consumption | Continuous leak versus legitimate load | Compare building automation/controller records and branch/submeter readings | [S4]; Anvilfield interpretation |
| Pressure falls during isolated test | Pressure changed in that boundary | Water loss, temperature change, valve behavior and stored volume effects | Document isolated-section investigation using the appropriate procedure | Hydrostatic testing; thermal expansion |
| Active flooding, loss of support or water near energized equipment | Damage requires immediate control | Full extent of damage | Stop testing; use safe authorized isolation and utility/emergency response; never enter potentially energized water | [S17]; Anvilfield response priority |
EPA's November 2023 commercial guidance describes listening equipment and leak-detection methods for concealed faults. After branch localization, select the acoustic or other locating method from pipe material, background sound and access rather than assuming a universal locating distance or depth. [S4]
CDC's August 24, 2026 flooded-home guidance permits power shutoff only from a dry location and calls for an electrician when reaching the switch requires entering standing water. Stop the meter test when water threatens energized equipment or structural support; arrange emergency control from a safe location. [S17]
How do you verify repair and restore service?
Repeat the original no-use comparison after repair using the same meter, unit and elapsed interval, comparable pressure and documented branch states. Allow refill and stabilization to finish first. Photograph both readings and report registered change, display increment and direction behavior instead of writing only “passed.”
Restore the repaired branch to its normal operating state and test again. An intermittent defect needs the next relevant operating cycle or interval data covering that cycle. Repeat the tank dye check for a tank-to-bowl repair; that test does not diagnose flushometer internals. [S2]
Restore valves, schedules and equipment in their required order; confirm cold and hot service and inspect the repair at operating pressure. Follow the installed heater's refill and restart procedure and close out any fire-system impairment under local procedures. [S16] [S18]
Badger Meter's November 2026 E-Series G2 manual warns that a continuous-flow alarm can take up to 30 minutes to clear after shutoff and directs the operator to check actual flow rather than only the alarm icon. Retain direct readings separately from notification timestamps. [S10]
How much water and money does the leak waste?
Anvilfield estimates that a sustained 0.4-gpm daily average wastes 576 gallons per day and 17,280 gallons over 30 days, about $134 in variable water charges at Seattle's 2026 inside-city off-peak rate. That rate is $5.82/CCF during September 16 through May 15. Water commodity cost is separate from fixed charges, sewer billing, repairs and damage. [S9]
For intermittent loss, multiply registered volume per episode by verified episodes per day. Anvilfield estimates 72 gallons per day for a 3-gallon refill occurring 24 times daily; keep that event count beside the estimate. Log repeated events to establish frequency.
| Constant flow | Gallons/day | Gallons/30 days | Water commodity cost/30 days, nearest dollar |
|---|---|---|---|
| 0.01 gpm | 14.4 | 432 | $3 |
| 0.1 gpm | 144 | 4,320 | $34 |
| 0.4 gpm | 576 | 17,280 | $134 |
| 1 gpm | 1,440 | 43,200 | $336 |
Cost method: convert lost gallons to CCF and multiply by the marginal local water rate. For a constant rate, 30-day water cost = Q × 43,200 / 748 × marginal $/CCF. The table uses Seattle's 2026 rate and 748 gallons/CCF, with total costs rounded to whole dollars; a cost row does not establish a meter's ability to measure that flow. [S9]
Seattle's inside-city residential peak tiers range from $5.98 to $11.80/CCF, so another season or billing tier changes avoided cost. Apply the address's actual sewer policy separately rather than presuming underground leakage earns a sewer credit. A small hidden leak can coexist with normal pressure; use pressure diagnosis for that separate symptom. [S9]
What belongs on the field checklist and utility record?
Portland Water Bureau's current leak-adjustment policy requires reasonable efforts to locate leakage and initiate repair within 30 days of the City's or customer's initial notification of increased usage, with completion within 90 days after the customer was notified. Each account can receive one qualifying underground/outdoor non-sewer adjustment and one sewer-entering leak adjustment per 12-month period, after repair and City meter verification. Apply the address's utility policy. [S12]
Keep discovery and repair dates, the leak description, repair explanation and verification readings for the owner's application and repeat visits. [S12]
Want this checklist to run itself on every job — with photo proof and a signed record crews can hand the customer? That's FieldOS.
How should facilities and public plumbing bids scope the work?
A diagnostic scope should deliver a verified meter and boundary map, no-use and branch logs, calculated rates with meter limitations, a supported active segment, the locating scope and repair-verification record. Assign meter-history retrieval, utility operations and schedule restoration.
Anvilfield's seven-state worksheet takes four hours including transitions, before repair. Additional branches, inaccessible valves, utility attendance and return visits for automatic cycles add labor and outage coordination.
Price access, occupied-building coordination, interval-data retrieval, specialty locating, repair and reinstatement separately. Schools and other public buildings need an agreed outage window and essential-load plan; those constraints often determine working hours. Public-building estimators use PorchRocket Bid Control to find plumbing solicitations and early service-line project signals in agency budgets and capital plans.
Define separate deliverables for investigation, locating, excavation and repair verification.
Review date and citation
Reviewed November 5, 2026. Sources list utility procedures, local policy and manufacturer editions; undated webpages are identified as current guidance.
Update log, 2026-11-05: Rechecked utility and meter specifications; separated reconnection filling from the intermittent-refill example; clarified valve authority, maintenance boundaries, heater protection and emergency response.
How to cite this page: Anvilfield, Water-meter isolation tests for hidden leaks: a utility-based field guide, 2026, https://anvilfield.com/field-guides/plumbing/water-meter-isolation-hidden-leak-diagnosis/
FAQ
How do water-meter isolation tests find hidden leaks?
Compare no-use readings across verified valve boundaries, then repeat the suspected state to identify the segment carrying unexplained flow. The fixture sequence and worked worksheet show the decisions.
What does a moving meter leak indicator mean?
It shows registered water passage. Exclude occupant use and automatic cycles before assigning that passage to leakage; a stored alarm describes history rather than necessarily showing current flow.
How long should I watch the meter for a leak?
Seattle uses 30 minutes, EPA's November 2023 commercial guide one hour and EPA home guidance two hours. Extend observation for coarse registers or automatic cycles.
How do I calculate leak rate from meter readings?
Divide the change in gallons by elapsed minutes for gpm; multiply by 1,440 for a constant daily projection. Keep transition water outside comparison intervals.
Does a stationary meter rule out a hidden leak?
No detectable change occurred during that window. Low-flow under-registration, display resolution, reverse flow and an event outside the window remain unresolved.
Why does the meter run with the house shutoff closed?
A passing shutoff, another live metered branch or leakage in the remaining meter-to-valve piping explains continued registration. Prove closure and map upstream tees before assigning a buried service leak.
What is the difference between a continuous and intermittent leak flag?
Neptune's September 2021 ProCoder FAQ classifies use in all 96 daily 15-minute intervals as continuous and use in 50 to 95 as intermittent. Other meters use different algorithms. Those usage-bin flags neither prove a defect nor capture every leaking fixture.
Why is the leak symbol still on after I shut the water off?
Badger's November 2026 G2 manual allows up to 30 minutes for a continuous-flow alarm to clear. Check actual flow separately.
Can a leaking toilet pass a short meter test?
Yes. Tank water can escape between refills while the meter remains still. EPA WaterSense's 10-minute tank dye test detects silent tank-to-bowl leakage. Exclude the normal refill immediately after reopening its stop before calling a later episode a recurring leak.
Can a sewer camera locate a pressurized water-service leak?
Sewer CCTV inspects drain or sewer interiors. Meter isolation and supply-side locating address pressurized supply pipes; use the sewer camera guide for the separate drain investigation.
What is the smallest leak a water meter can detect?
Badger's November 2026 5/8-inch G2 displays 0.01 gallon and specifies a 0.04-gpm minimum test-flow point at ±3% for new meters at 45 to 122 °F. That point is not a starting-flow guarantee; use the installed model's body rating and display increment together.
Will the meter show a leak before the meter or on a private well?
Water escaping before the meter never crosses its measuring element. A private well without a service meter needs a suitable flow-measurement boundary or separate documented supply investigation; pump cycling alone does not quantify leakage.
People also ask
Sources checked
- Seattle Public Utilities, Detect Leaks, current undated web page 30-minute no-use comparison and building shutoff check. Checked 2026-11-05.
- EPA WaterSense, Home Maintenance, current web page Two-hour home meter check; annual silent-toilet check; 10-minute dye procedure. Checked 2026-11-05.
- Portland Water Bureau, How to check for leaks, current web page Indicator types; several-hour slow-leak observation; verify main-valve closure at faucet; failed isolation prevents location inference; irrigation/appliance suspects. Checked 2026-11-05.
- EPA WaterSense at Work, Section 2.2, Leak Detection and Repair, November 2023 One-hour commercial no-use comparison, pp. 2 onward; AMI reporting intervals, submeters, visual/auditory checks and listening methods. Checked 2026-11-05.
- Neptune Technology Group, ProCoder Frequently Asked Questions, FAQ PROCODER 09.21 Eight-wheel bank versus finer sweep hand; 96 fifteen-minute intervals; 50 to 95 intermittent flag classification; endpoint-dependent feature access. Checked 2026-11-05.
- Neptune Technology Group, ProCoder)R900i product sheet, 22-007174, 02.22 Size- and unit-specific register increments, p. 2; 15-minute flags and 96 days of hourly history. Checked 2026-11-05.
- Neptune Technology Group, T-10 Meter, 1-1/2 and 2 inch, 23-010382, 07.23 Normal and low-flow body performance, p. 3. Checked 2026-11-05.
- Portland Water District, How do I read my residential water meter?, current undated web page Billing HCF conversion, 748 gallons/HCF; utility example 900 minus 897 = 3 HCF = 2,244 gallons; stop ice maker/toilet use during meter check. This is Maine's district, separate from Portland, Oregon. Checked 2026-11-05.
- Seattle Public Utilities, Residential Drinking Water Rates, 2026 schedule effective through December 31, 2026 748 gallons/CCF; $5.82/CCF inside-city off-peak commodity rate; seasonal dates and peak tiers; input to original water-cost estimates. Checked 2026-11-05.
- Badger Meter, E-Series G2 lead-free bronze meters, 5/8 to 12 inch, User Manual ESM-UM-04284-EN-13, November 2026 Totalizer increments and net volume, pp. 15-16; endpoint resolution, p. 19; suspected leakage, p. 22; performance and temperature conditions, p. 24; alarm clearing, p. 28. Manufacturer index lists revision modification date 2026-11-03. Checked 2026-11-05.
- Badger Meter, E-Series Cold Water Engineered Polymer, Product Data Sheet ESM-DS-00041-EN-09, November 2022 Engineered-polymer meter increments, minimum test flows and normal ranges, p. 2. Checked 2026-11-05.
- Portland Water Bureau, Request a bill adjustment due to water leak, current web policy summary Repair initiation within 30 days, completion within 90 days, two category-specific adjustment allowances per 12 months; city meter verification and required record fields. Checked 2026-11-05.
- City of Portland Code, Chapter 21.30 Water Services, current online text Sections 21.30.130, 21.30.200 and 21.30.320: local maintenance/repair boundaries and required backflow protection. Checked 2026-11-05.
- Badger Meter, E-Series Residential Ultrasonic Flow Meters, current manufacturer document index Manufacturer document index and product-family distinctions. Checked 2026-11-05.
- Neptune Technology Group, T-10 Meter, sizes 5/8, 3/4 and 1 inch, 23-010383, 07.23 Small T-10 body normal ranges and 95% low-flow registration points, p. 4. Checked 2026-11-05.
- A. O. Smith, Residential Condensing Gas Water Heaters, Series 120-121, 100368549_2000624920B, printed November 2023 Full-tank operation and open cold supply, p. 45; unobstructed temperature-pressure relief valve and discharge, p. 37. Checked 2026-11-05.
- CDC, Safety Guidelines: Reentering Your Flooded Home, August 24, 2026 Electrical isolation from a dry location; electrician needed if access requires entering standing water; structural-damage entry precautions. Checked 2026-11-05.
- Seattle Fire Department, Systems Testing, current web guidance Fire-system impairment notification, mitigation and restoration; local procedures remain separate from domestic leak testing. Checked 2026-11-05.
- Seattle Public Utilities, How to Shut Off Your Water, current web guidance Building-valve preference; customers may operate top shut-off meters; older meter valves require SPU crews. Checked 2026-11-05.