Investigation field dictionary

Intermittent fault and no-fault-found data fields

Separate observable events, operating context, service outcomes, test exposure, and conclusions so one field is not asked to prove more than it records.

Event identity and observation

FieldDefinition and boundary
Asset or unit IDStable physical tag or serial identity; do not substitute a temporary location alone.
Event timeTimestamp with a declared source and time basis; record synchronization uncertainty.
Observable symptomWhat changed or was detected, without embedding an unverified cause.
Operating phaseControlled state such as startup, acceleration, steady load, transition, shutdown, or idle.
DurationObserved or estimated elapsed time with unit and estimation method.
Fault code or alarmNative displayed or logged value; preserve the original source alongside normalized text.
RecoveryObserved return mechanism such as self-clear, reset, power cycle, swap, or still present.
Intervention sequenceOrdered actions after the event; distinguish evidence captured before and after each action.

Context and source evidence

FieldDefinition and boundary
Load / operating stateRelevant command, output, duty, mode, and adjacent-system state at the event.
EnvironmentMeasured or observed temperature, moisture, contamination, vibration, or other declared condition.
ConfigurationHardware, software, firmware, parameter, and installation state needed to compare events.
Original evidenceNative log, waveform, trend, photo, video, or witness source with retention location and owner.
Evidence gapA declared missing source; absence must not be silently converted to normal or not applicable.

Service-history ratios

Target-unit event concentration (%) = target-unit events ÷ all comparable events × 100

The population must be a consistent peer group and period. The ratio is not exposure-normalized unless each unit had equal opportunity.

No-fault-found share (%) = NFF shop outcomes ÷ comparable shop returns × 100

Define NFF from the actual shop process and test coverage. It does not mean “proven serviceable in all operating conditions.”

Symptom recurrence after swap (%) = cases with repeated declared symptom ÷ swaps with adequate follow-up × 100

Keep the symptom definition and follow-up window stable. Recurrence does not by itself locate the fault.

Pattern and reproduction fields

FieldDefinition and boundary
Repeated signatureAn exact match of normalized phase, code, and recovery in the ReliabilityBench screen; it is an observation cluster, not a causal model.
BaselinePreserved as-found or approved known-good state from which intentional changes are compared.
TriggerOne evidence-based condition deliberately staged within approved limits.
InstrumentationChannels, range, sample rate, synchronization, trigger logic, and retained source format.
ExposureRepetitions, dwell, cycles, starts, hours, or other opportunity actually tested.
Stop criteriaPreapproved condition requiring the stage to stop or hold.
Negative resultTarget symptom not observed within the declared conditions and exposure; never equivalent to universal absence.
RestorationVerified return to the approved baseline and separate release-to-service process.

Decision boundary

Use these fields with approved manuals, safety and energy controls, configuration management, source-data retention, qualified engineering judgment, and accountable release authority. The linked workflow tools organize records; they do not authorize testing or diagnose a cause.