Definition
Ratio of actual radiation dose accumulated during an AI-scheduled work sequence to the dose predicted by the ALARA plan for that sequence, measuring whether the scheduler's task ordering creates dose exposures the ALARA plan didn't anticipate.
The Measurement Gap
AI-optimized maintenance schedules in radiological areas reorder tasks to maximize resource efficiency, but the task sequence directly affects the dose-rate exposure profile for the work crew. ALARA plans are developed with a specific task sequence assumption. When the AI scheduler reorders the sequence without coordinating with the ALARA plan, the actual dose may differ from the planned dose even if total work scope is unchanged. No legacy metric measures the alignment between the AI-scheduled task sequence and the ALARA plan's dose assumptions. CRE (collective radiation exposure) measures total dose outcome. ALARA plan compliance tracks whether the ALARA plan was followed. Schedule compliance tracks whether work was done on time. None of them measure whether the scheduling sequence itself created a dose profile different from what the ALARA plan assumed. The invisible failure mode is that the scheduler reorders tasks in a way that places the crew in higher dose-rate areas for longer periods than the ALARA plan anticipated, and the dose variance is attributed to field conditions rather than the scheduling decision.
Decision Supported
Is the AI-optimized maintenance sequence in radiological areas aligned with the ALARA plan's dose assumptions, or is the scheduler reordering tasks in ways that change the dose-rate exposure profile without updating the dose prediction?
Formula
Plain language: For each AI-scheduled work sequence in a radiological area, compare the actual collective dose accumulated during execution to the dose estimate in the ALARA plan. A ratio above 1.0 means the sequence produced more dose than planned. Decompose the variance into sequence-attributable factors (task ordering changed the dose-rate profile) vs. other factors (actual dose rates differed from estimates, work took longer than planned).
Notation:
SDAF = Dose_actual / Dose_ALARA_planned
Sequence attribution:
SDAF_sequence = Dose_actual / Dose_if_ALARA_sequence
Variables:
- Dose_actual = actual collective radiation dose accumulated during the AI-scheduled work sequence (person-mrem or person-μSv)
- Dose_ALARA_planned = dose estimate from the ALARA plan for the planned sequence (same units)
- Dose_if_ALARA_sequence = estimated dose if the ALARA plan's original sequence had been followed with actual dose rates and actual work durations (same units: isolates the sequence effect from the dose rate and duration effects)
- SDAF captures all sources of dose variance; SDAF_sequence isolates the AI scheduling contribution
Sign convention: SDAF = 1.0 means dose matched the ALARA plan. Above 1.0 means more dose than planned (bad). Below 1.0 means less dose than planned (good: the AI sequence was better than the ALARA plan assumed). Focus attention on SDAF > 1.0 cases.
Aggregation period: Per radiological work evolution, rolled up per outage and per rolling 12-month period.
Edge Cases
- ALARA plan was overly conservative (estimated more dose than any realistic sequence would produce): SDAF will be below 1.0 regardless of the schedule quality. Compare SDAF trend over time rather than absolute value.
- Work sequence changed mid-execution due to field conditions: evaluate both as-scheduled and as-executed SDAF. The scheduling decision may have been sound but the field changed it
- Very low dose evolutions (< 10 person-mrem): exclude from aggregate SDAF. Small absolute doses produce noisy ratios
- Minimum: all radiological work evolutions above 100 person-mrem planned dose
Proposed formula, pending validation pilots.