A sealed enclosure that failed leak testing. The enclosure was an aluminum gearbox housing, 2 liters internal volume, rated for 0.5 bar pressure. The customer tested it with a pressure decay test: fill to 0.5 bar, wait 30 seconds, measure the pressure drop. The allowable drop was 0.5 mbar. The housing failed with 2 mbar drop. The customer thought there was a crack. They pressure-dyed it. No crack. The issue was the test method — the housing was leaking, but the decay test couldn’t distinguish between a leak and thermal expansion. This is about leak testing methods and why pressure decay needs temperature compensation.
The pressure decay test
A pressure decay test fills the part to a set pressure, isolates it from the supply, and measures the pressure drop over time. A drop indicates a leak. The calculation:
Q = (ΔP / P) × V / t
Where ΔP is the pressure drop (Pa), P is the absolute pressure (Pa), V is the part volume (m³), and t is the test time (s). For a 2 L housing at 0.5 bar gauge (1.5 bar absolute), 2 mbar drop in 30 seconds: Q = (200/150000) × 0.002 / 30 = 0.00133 × 6.67×10⁻⁵ = 8.9×10⁻⁸ m³/s. Converting to sccm: 5.3 sccm. That’s a small leak — about a 0.1 mm hole. But the customer’s spec was 0.5 mbar drop (1.3 sccm). The 2 mbar was 4x the spec.
The problem: when the housing was filled with compressed air, the air warmed up (adiabatic compression). After 30 seconds, the air cooled to ambient. The pressure dropped 1.5 mbar purely from temperature change. The actual leak was only 0.5 mbar (within spec). The thermal effect masked the test. The housing was good, but the test said it was bad.
What was changed
1. Added a stabilization time. After filling, the test waits 10 seconds before starting the measurement. The air cools to ambient during stabilization. The pressure settles. The measurement starts from a stable baseline. The thermal expansion is excluded. The 2 mbar drop became 0.3 mbar (within spec). The housing passed.
2. Used a master part comparison. A known-good housing was tested alongside the production part. The tester compared the production drop to the master drop. Any difference indicated a real leak (the thermal effects cancel out between the two parts). The master test is more accurate than absolute decay. It accounts for temperature, volume, and test time.
3. Switched to mass flow measurement. For high-volume production, a mass flow meter measures the air flow needed to maintain the test pressure. A leaking part needs continuous make-up air. The flow rate is directly measured (no calculation). The test is faster (no 30-second wait) and more accurate. The mass flow system costs $5000 but tests 1000 parts per hour. The pressure decay system tests 60 parts per hour. For production volumes, the mass flow pays for itself.
The leak test methods comparison
| Method | Sensitivity | Cycle time | Cost | Use for |
|---|---|---|---|---|
| Underwater bubble | 0.1 sccm | 60 s | $100 | Prototype, visual leak location |
| Pressure decay | 1 sccm | 30 s | $1000 | Production, simple parts |
| Differential decay | 0.1 sccm | 30 s | $3000 | Precision, small leaks |
| Mass flow | 0.01 sccm | 5 s | $5000 | High-volume production |
| Helium leak test | 10⁻⁶ sccm | 120 s | $20,000 | Medical, aerospace |
The test pressure choice
The test pressure should be 1.5x the working pressure. A gearbox rated for 0.5 bar internal pressure is tested at 0.75 bar. Too low a test pressure won’t reveal a small leak. Too high and the part distorts (changing the volume and the baseline). The test pressure must not stress the part. For a gearbox at 0.5 bar working, 0.75 bar test is standard. For a sealed enclosure at 0.1 bar working (watertight), 0.2 bar test is enough. The test time depends on the leak rate: smaller leaks need longer test times. A 30-second test at 0.5 bar detects 1 sccm. A 5-minute test detects 0.1 sccm.
The leak test rule: always add a stabilization time after filling. The failed housing wasn’t leaking — 1.5 mbar of the 2 mbar drop was thermal cooling. Stabilize 10 seconds, then measure. Use a master part for comparison. For high-volume production, mass flow is faster and more accurate. Choose test pressure 1.5x working. The test method determines the rejection rate — a bad test rejects good parts.
