Weather and outage stress ========================= The operational stress test applies an assumed weather and outage calendar to the median rolling policy's first-year shift package. It is a feasibility screen, not a forecast, outage schedule, or safety certification. Assumptions ----------- - 75-day fall work window. - Fixed 5-day outage block. - Independent 20% weather-loss probability on each day in the window. - 50 scenarios, seed 7. - The baseline uses the same 75-day window with no blocked days. Network result -------------- Baseline: feasible, 116 shift assignments, 813.7 productive hours, 8.5 travel hours, 63 days used. Stress: - fully feasible in **0/50 scenarios**; - mean work completion: **93.4%**; - median unscheduled work: **50.2 hours**; - P90 unscheduled work: **97.7 hours**; - mean blocked days: **19.4**; - median shift assignments: **107**. The assumed calendar removes enough field days that no sampled scenario schedules every work item. The package completes most hours but leaves a median 50 hours unscheduled. This is a capacity and calendar finding, not evidence that the policy is field-infeasible. Interpretation -------------- The first-year policy package was feasible with no weather or outage closures. Under the assumed calendar, it is not fully schedulable. The likely remedies are a longer window, additional crews or machines, a smaller first-year scope, or a policy that reserves weather slack. Those are design choices, not automatic conclusions. Limits ------ - Weather loss is an independent per-day assumption, not a forecast or field record. - The fixed outage block is assumed, not an asset-owner outage schedule. - Only the median policy episode and its first year are stressed. - No field calibration, safety certification, or clearance standard is claimed. All-episode scan ---------------- The calendar stress was repeated for all 20 rolling episodes, with 50 scenarios per episode. .. list-table:: :header-rows: 1 * - Metric - Value * - Baseline 75-day feasible - 50.0% * - Scenario-level full completion - 13.2% * - Episodes with full stress completion - 2 of 20 * - Mean work completion - 85.6% * - Unscheduled hours P50 - 127.5 * - Unscheduled hours P90 - 372.3 Failure sources across the 20 episodes: - fall-window capacity: 10 episodes; - weather/outage calendar: 8 episodes; - crew capacity: 0 episodes; - fully feasible: 2 episodes. The 75-day window is the primary constraint. Increasing the window improved scenario-level full completion to 40.2% at 90 days and 62.8% at 105 days, with mean unscheduled work falling from 165.8 to 83.7 and 38.3 hours. A longer window alone does not reach full completion under the assumed weather calendar.