The Seal That Leaked Only in the Field
A small enclosure for an outdoor sensor uses a molded gasket compressed by six M3 screws into a plastic housing. Bench testing with a 0.5 psi air-decay leak test passed 50 of 50 units at 20 °C. In the first field winter, 8% of 2,000 deployed units reported moisture ingress. Field temperatures reached −25 °C. The gasket is a general-purpose EPDM, 2.0 mm nominal, compressed about 25%. The housing is glass-filled polyamide. Screws are tightened to a torque spec with a calibrated driver. A customer asks for a root-cause report and a corrective action within 30 days. The team's first instinct is to increase screw torque and add a second gasket.
1. Critique the instinct to increase torque and add a gasket. What is the more likely mechanism?
Read the worked answer
Both instincts treat the leak as insufficient squeeze, but the bench data argue against that: 50 of 50 passed at 20 °C with the current squeeze, so the room-temperature seal is adequate. The failure appeared only in cold field conditions, which points to a temperature-driven mechanism, not a torque shortfall.
Likely mechanism: differential thermal contraction. EPDM and glass-filled polyamide have different coefficients of thermal expansion, and at −25 °C the housing and gasket shrink differently, reducing compression or causing the gasket to lose contact at corners or between screws. EPDM also stiffens in cold, reducing its ability to conform and recover. A general-purpose EPDM may also be near or past its low-temperature flexibility limit.
Increasing torque is risky: it may over-stress the plastic bosses, cause creep, or crack the housing, and it does not address differential contraction. Adding a second gasket adds a joint and cost without a mechanism.
Better direction: verify compression at −25 °C (measure gap or use pressure-sensitive film), check the gasket's cold flexibility rating against −25 °C, and evaluate a colder-rated material or a design that maintains squeeze across the temperature range.
2. Design a verification that would reproduce the field failure before any fix is released.
Read the worked answer
The bench test must include the missing variable: cold. Reproduce the field condition, not just the lab condition.
Test: take production-representative units, condition them at −25 °C for a soak long enough for the whole assembly to reach temperature (hours, not minutes), then run the same 0.5 psi air-decay leak test while cold, and again after cycling back to room temperature. Include thermal cycling (−25 to +50 °C) to capture any set or creep.
Sample: enough units to detect an 8% rate with reasonable confidence - 50 units gives a decent chance of seeing several failures; fewer units may miss it.
Instrumentation: measure actual gasket compression at cold using shim or film at corners and mid-span, since corner and between-screw locations are the likely leak paths. Record screw torque before and after cycling to detect relaxation.
Pass criteria: zero leakage cold and after cycling, with measured compression staying above the minimum needed for sealing across the range. If the current design fails this test, the test is validated as a reproduction and becomes the acceptance test for the fix.
3. The 30-day deadline is firm. Propose a corrective action and state its residual risk.
Read the worked answer
Corrective action should address the mechanism: loss of compression and cold stiffness.
Proposed fix: (1) change the gasket to a material rated well below −25 °C (e.g., a low-temperature EPDM or silicone compound with documented cold flexibility) and verify compression set; (2) increase the number of screws or add corner fasteners so compression is maintained between fasteners where contraction is worst; (3) add a compression limiter or controlled boss geometry so torque produces consistent squeeze without over-stressing plastic; (4) if geometry allows, increase nominal squeeze modestly to absorb differential contraction.
Validation: the cold leak test and thermal cycling above, on at least 50 units, before release.
Residual risk to state: the fix is validated over a limited number of cycles and one cold extreme; long-term compression set and UV/ozone aging of the new material are not fully covered in 30 days. Recommend an ongoing aging test and a field monitoring plan, and be explicit that 50 units cannot prove a rate below a few percent. Report the mechanism, the evidence, the fix, the validation, and the residual risk - that is what the customer asked for.