The Repeated Small Release and the Investigation That Wasn't
A chemical site has had three small releases of a corrosive liquid from a pump seal over eight months. Each was cleaned up by the shift team, logged as a maintenance defect, and repaired. No incident investigation was raised because each release was under the site's reporting threshold (5 litres). The third release sprayed a fitter's forearm; he was wearing goggles but not a face shield and received a minor burn. The pump is in a congested area with limited access. The site's incident procedure defines reportable incidents by volume and injury severity only. A recent audit found that 60% of maintenance defects are closed without a root cause code.
1. Explain why treating these releases as non-events is a flawed approach, and what the pattern indicates.
Read the worked answer
The threshold-based approach misses the signal in the pattern:
- Recurrence is the signal: Three releases from the same pump seal in eight months is not random. It indicates a persistent failure - wrong seal material for the duty, misalignment, dry running, or a repair that does not address the cause. Each repair that 'fixes' the symptom without a root cause resets the clock.
- Threshold logic is backwards: Defining reportability by volume and injury severity means a near miss with high potential (spray to the face, corrosive liquid) is invisible. The fitter was one step from an eye injury. Potential consequence, not just actual outcome, should drive investigation.
- Defect closure without root cause: 60% of defects closed without a root cause code means the organisation cannot see patterns. The three seal failures were probably recorded as three unrelated repairs.
- Barrier weakness: The fitter wore goggles but not a face shield, so the PPE barrier was inadequate for a spray hazard from a pressurised seal - a sign the task risk assessment is out of date.
Conclusion: this is a recurring loss-of-containment precursor that should have triggered investigation after the first or second event.
2. Design an investigation approach for the pattern and the injury, and recommend changes to the incident and defect systems.
Read the worked answer
Investigation approach:
- Scope: investigate the pattern (three releases) and the injury together, not as separate minor events. Use a method suited to recurring failures (e.g. fault tree or barrier analysis) rather than a single-event timeline.
- Evidence: pump maintenance history, seal type and duty, alignment and vibration records, operating conditions, the task risk assessment, and the PPE actually specified for seal work.
- Questions: why did the seal fail repeatedly; why was each repair treated as isolated; why was no investigation raised; why was face protection not required for a spray hazard in a congested area.
- Findings and actions: correct the seal application (material/duty), improve access, update the task risk assessment and PPE, and add a verification step after repair.
System changes:
- Replace volume/severity-only reporting with a potential-consequence screen: any loss of containment of a corrosive, or any event with serious-injury potential, triggers investigation regardless of volume.
- Require a root cause code before a defect can be closed; audit the 60% closure gap and reopen significant defects.
- Add recurring-defect trending (e.g. same equipment, same failure mode) to the process safety review.
- Track investigation closure and action effectiveness, not just investigation counts.
Key principle: investigate for learning and for pattern detection, not only to satisfy a reporting threshold.