Definition
Percentage of degradation mechanisms required to be evaluated by the applicable ASME code inspection requirement that the AI visual inspection system is capable of detecting, measured per component examination category.
The Measurement Gap
AI-powered visual inspection systems in nuclear plants are qualified to detect specific types of degradation: cracking patterns, wall thinning, surface corrosion, weld defects. But ASME Section XI inspection requirements specify a full set of degradation mechanisms that must be evaluated for each component examination category. No legacy metric compares what the AI can detect against what the code requires it to evaluate. FME event rate tracks foreign material intrusion. PM completion rate tracks whether inspections are performed. WANO indicators track broad inspection program health. None of them measure whether the AI inspection system covers the full scope of degradation mechanisms the applicable code requires. The invisible failure mode is that the AI inspection program is assumed to provide complete coverage because it produces inspection results, but it may detect only a subset of the required degradation mechanisms while the program treats the results as if full-scope inspection was performed.
Decision Supported
Does the AI inspection system cover the full scope of what the code requires, or does it detect a subset of mechanisms while the program assumes full coverage?
Formula
Plain language: For each component examination category (e.g., VT-1 on reactor vessel welds, VT-3 on containment surfaces, UT on piping), list the degradation mechanisms that the code and plant-specific augmented inspection plan require the inspection to evaluate. Determine which of those mechanisms the AI system can reliably detect. Divide detectable by required.
Notation:
IDMCR_c = DM_detectable_c / DM_required_c
Plant composite:
IDMCR = Σ_c (w_c × IDMCR_c) / Σ_c w_c
Variables:
- DM_detectable_c = number of degradation mechanisms in examination category c that the AI inspection system has demonstrated capability to detect (count)
- DM_required_c = total number of degradation mechanisms in examination category c that the code/augmented inspection plan requires evaluation for (count)
- w_c = safety significance weight of examination category c (based on component safety classification and consequence of mechanism: 1.0 for ASME Class 1, 0.7 for Class 2, 0.4 for Class 3, 0.2 for non-code)
- c = examination category (e.g., VT-1 welds, VT-1 bolting, VT-3 surfaces, UT volumetric, ISI piping)
Sign convention: IDMCR = 1.0 means the AI covers every required mechanism. Lower is worse. A value of 0.6 means the AI misses 40% of what the code requires.
Aggregation period: Per outage cycle (updated when AI inspection capabilities are qualified or when the code/inspection plan changes).
Edge Cases
- New degradation mechanism identified (e.g., a new form of environmentally-assisted cracking): IDMCR_c drops immediately for the affected category, flagging a gap before the next inspection
- AI detects a mechanism but with lower probability of detection (POD) than the code requires: count as partially detectable (0.5) rather than binary (1 or 0) if POD is documented. If POD is unknown for the mechanism, count as undetectable until qualified.
- Multiple AI systems cover different mechanisms: union of all AI detection capabilities counts toward DM_detectable
Proposed formula, pending validation pilots.