Gas Leak on the Production Line: How to Detect and Stop It Before Appliances Ship

Where a Production-Line Gas Leak Comes From

Leak Detection Methods Compared

Pressure decay is the production-line standard because it gives a pass/fail value in seconds, is independent of the operator, and can be logged. Bubble testing is excellent for finding exactly where a failed unit leaks, but it is too slow and subjective to gate every unit. Pressure should be held long enough to reveal small leaks; an over-short hold time passes units that fail later.

How a Pressure-Decay Test Runs

1. Connect the bench to the appliance inlet and seal the downstream outlet(s);
2. Pressurise the circuit to the test pressure using clean, dry air or nitrogen — never the combustible gas during a leak gate;
3. Allow a short stabilisation time so the pressure reading settles;
4. Hold for the defined time and measure the pressure drop;
5. Pass if the decay is within the limit set by the standard and the part; fail and divert to repair if it exceeds it.

Two tightness checks are usually required: external tightness (no gas escapes to the room from joints, tubes and valve bodies) and internal tightness (a closed valve does not pass gas to a downstream burner). A valve that closes on an FFD command but still leaks internally is not safe, which is why the leak gate and the FFD test (F1) share the same record. Temperature, volume and stabilisation time all affect a pressure-decay reading, so the bench must use a consistent volume and hold time; otherwise a genuine small leak can be masked by pressure settling.

Cycle Time, Layout and the Repair Loop

• To keep takt, the fill/stabilise/hold sequence is sized to the line cycle; slow or large-volume circuits can use a dedicated station rather than slowing the whole line.
• Failures are diverted to a repair bay, located with bubble solution or a sniffer, corrected, and then re-tested from the start — a repaired unit is not assumed tight.
• The tightness result, pressure-decay value and rework status are stored against the serial number.
• Repeated leaks at the same joint or component reveal a torque-tool, jig or supplier problem to fix upstream, rather than individual rework.

An automated gas cooker test bench or gas appliance test bench performs the fill, hold, measure and pass/fail decision, opens the interlock only for tight units, and logs the decay value — removing the operator judgement that lets a slow leak reach the customer.

Calibration, Leak Limits and Why 100% Testing Pays

A pressure-decay reading is only as trustworthy as the pressure reference and the test volume, so the bench’s pressure transducer must be calibrated on a schedule and the test volume and connection kept identical between units. The pass limit must come from the standard and the product, not from a number the factory finds convenient; the bench should convert the measured decay to a leak rate so that a slightly different circuit volume still gives a comparable result. A common failure mode is a limit set from a “good” sample rather than from the regulation — a drift that quietly passes marginal units.

The reason this matters is cost. A gas leak found at the line is a few minutes of repair in the existing bay. A unit that passes the factory and is found leaking after export is a warranty claim, a return shipment, a possible rejection of the whole container at customs, and a field-safety record that can lead to withdrawal. Sampling leaves the leak rate to chance: if a joint fails at a rate of, say, a few units per thousand, even a large sample can miss it while every untested unit is shipped. A 100% gate costs a few seconds per appliance and removes that probability entirely, which is why tightness is treated as a safety gate rather than an audit item. The stored decay values also protect the manufacturer in a dispute: when a field unit is alleged to have leaked, the serial record shows whether it passed the external and internal tightness checks at the point of manufacture. In practice, factories that keep these records consistently report fewer field leaks and far shorter arguments with importers, because the evidence is already on file rather than reconstructed after the fact.

Common Mistakes

• Using bubble testing as the main gate instead of an objective 100% pressure-decay test;
• Hold times too short to reveal small leaks, or no stabilisation step;
• Testing only external tightness and missing internal valve leakage;
• Running the leak test with the combustible gas rather than air or nitrogen;
• Releasing a repaired unit without a full re-test;
• Recording only pass/fail, so a recurring leak point cannot be traced;
• Not interlocking the leak gate with the FFD and the line’s pass system.