Failure Mode Record

This is the classification step, not the repair step and not the investigation. It sits after the asset is fixed and before any decision about changing maintenance strategy. Using it for neighbouring jobs produces records nobody can report on.

ISO 14224 is a data collection standard rather than a management system standard: it prescribes what a failure record must contain and how it is classified, not certification. The mapping below shows where its requirements land on this form.

ISO 14224 carries no legal force. What varies is whether a regulator or customer requires documented evidence of equipment condition and deficiency, at which point a coded failure history stops being an engineering nicety.

Filling the form is quick. Whether the record is worth anything a year later turns on four judgement calls, each of which a busy technician gets wrong by default unless the expectation is set.

These are what show up when a coded failure programme is audited a year after launch. All are visible in the data itself, which is the point.

A dairy processor logged twenty-three unplanned stops on one homogeniser across two years. Every close-out said some version of "bearing failure, replaced, tested, returned to service". Bearings duly topped the plant's failure report, so the response was a premium bearing brand and a shorter replacement interval. Failures continued at the same rate at greater cost.

Coded records changed the picture in four months. Component Failed was indeed Bearing, but Failure Mechanism came back as Corrosion and Pitting rather than fatigue, Environment Contributing was answered Significantly on nine of eleven records, and Detection Method was Failure in service every time. The bearings were not wearing out, they were being washed. The fix was a shaft seal redesign and a change to the caustic wash routine, neither of which had ever been proposed while the record was prose.

Coding one failure takes two minutes. Making sure every failure gets coded, that codes stay consistent between shifts, and that patterns reach someone who can act is the part that decays quietly.

KnowMaintain

Holds the asset register the record codes against, links each failure to its work order and case thread, and keeps the coded history on the asset rather than on the job.

KnowQuality

Picks up the CAPA when Action Required is Yes, so the corrective action lives with every other nonconformity rather than in a maintenance spreadsheet.

KnowSafe

Catches failures with a safety consequence, particularly guard, interlock and control failures, and routes them into incident and risk review.

Ella

Watches the coded population for repeats, flags assets crossing a bad actor threshold, and drafts the review trigger for your approval before any write.

Maintainable item

The lowest level in the ISO 14224 taxonomy at which a failure is recorded and a repair carried out, typically the bearing or seal rather than the pump.

P-F interval

The time between an incipient failure becoming detectable and functional failure. It sets the maximum useful interval for a condition monitoring route.

Bad actor

An asset or component whose failure frequency or consequence places it in the small population accounting for most unplanned downtime.

Infant mortality

Failure concentrated just after installation or overhaul, usually caused by fitting, commissioning or specification rather than wear.

MTBF

Mean time between failures, the average operating interval between unplanned failures of a repairable item. Meaningful only when the population is consistently coded.