A Precision Line Test Fails Three Weeks After a Clean Tank Test
During an annual compliance test at a PEI retail site, the precision line tightness test on a pressurised product line (0.38 L/h threshold) records an apparent loss of 0.55 L/h over the one-hour test period. The tank tightness test passed three weeks earlier. The test was started 35 minutes after a 15,000 L delivery; the product had warmed about 4 °C while the delivery was going in. The line is 45 m of 50 mm inside-diameter flexible pipe, roughly 88 L of line volume. The mechanical line leak detector, rated at 3.0 gph, passes. There is no visible product in the STP sump or the dispenser sumps and no sheen in the monitoring wells. The contractor's report proposes declaring a failed line and excavating the entire run to replace the piping, at considerable cost and with the site shut down for several days. The owner has asked for a second opinion before authorising the work.
1. Before accepting a failed-line finding, what could produce this result, and what does the arithmetic suggest?
Read the worked answer
Do the arithmetic first - it points hard at thermal effects, not a leak.
Gasoline's coefficient of volumetric expansion is around 0.00095 per °C. Over an 88 L line, a 4 °C change gives roughly:
88 L × 0.00095 × 4 °C ≈ 0.33 L of contraction.
That is the same order of magnitude as the 0.55 L/h apparent loss. The product warmed 4 °C during the delivery and then cooled in the line; a cooling, contracting line looks exactly like a line losing product to a fixed-volume test. Add the fact that the test was started only 35 minutes after a 15,000 L delivery - the line had not thermally stabilised and may also have carried entrained air or vapour drawn in during the delivery.
Other candidate causes of a false failure:
- Line not fully primed or an air pocket compressing/absorbing during the test.
- Pressure stabilisation period too short before the measurement window began.
- A leaking STP check valve allowing slow backflow past the pump.
- Transducer or gauge drift, or a meter past calibration.
- Test method without temperature compensation, on a short line where the volume change is a large fraction of the signal.
Crucially: a 3.0 gph leak detector passing tells you nothing about a 0.55 L/h loss - it is nearly eight times less sensitive. It neither confirms nor refutes the precision test.
2. Design a defensible re-test protocol that would stand up to regulatory scrutiny.
Read the worked answer
The re-test has to remove each alternative explanation, one by one, or the result means nothing.
Pre-conditions
- No deliveries for a defined period before the test - a minimum of several hours, and preferably the site is left quiet overnight. Record the time of the last delivery.
- No dispensing during the stabilisation and test period; dispensers locked out and tagged.
- Confirm the line is fully primed, with the pump run and pressure stable before the measurement window starts.
- Allow a proper stabilisation period with pressure held, not 35 minutes after a delivery.
Measurement 5. Record product temperature at the start and end of the test with a calibrated probe, and at more than one point on the line. Use a test method with temperature compensation, or run the test in a way that the thermal drift is quantified and shown to be smaller than the threshold. 6. Use a calibrated, in-certification test instrument, and record the instrument's calibration date on the report. 7. Run the test at least twice, ideally on consecutive days, and report both results.
Concurrent observation 8. Sump and observation well checks immediately before and after each test, recorded with photos. 9. Inventory reconciliation over the test period.
Reporting A pass on the re-test, properly stabilised and temperature-corrected, is a legitimate result and closes the file. A repeat failure under these conditions is a real finding, and the excavation discussion becomes reasonable - but only then.
3. If the re-test confirms a genuine loss, how do you localise it without excavating the whole 45 m run?
Read the worked answer
Do not dig a 45 m trench to find one leak. Split the system and let the test tell you where to look.
Step 1 - the high-probability points. Statistically, the leak is at a connection, not mid-pipe: the STP discharge fitting and flange, the dispenser shear valve and flex connector, the sump entries, and any transition or transition sleeve. Inspect every one, dry, with the pump running under pressure, and look for a weep that only appears under pressure.
Step 2 - sectionalise. Close the shear valves at each dispenser and test in sections: pump to first tee, each dispenser branch individually, and the dispensers themselves. Isolating one branch at a time will bracket the loss to a short length of run within a couple of test cycles.
Step 3 - test the secondary containment independently. A vacuum or hydrostatic test on the secondary-contained annular space can confirm whether the primary pipe is breached, and can sometimes pinpoint the section, without exposing the primary pipe.
Step 4 - non-destructive localisation. Tracer gas or acoustic leak location over the suspect section, guided by the sectional results, before any excavation.
Step 5 - verify after repair. Any repair is followed by a full re-test to the same threshold and method as the original, plus a functional test of the line leak detector, and the whole sequence is documented.
One more thing: whatever the outcome, the sump finding, the test anomaly and the corrective action all go in the site's records, and any confirmed release is reported under the PEI Petroleum Storage Tank Regulations. A pass on a re-test does not erase the original result - the file must show why it was set aside.