Risk-ranked scope deferral ========================== The proportional scope remedy defers hours mechanically. This comparison instead defers whole first-year actions ranked by **lowest first-year risk reduction per hour**, until about 20% of action hours are deferred. This is first-year risk ranking, not a multi-year risk model. The actual deferred fraction can overshoot because whole actions are selected. Network result -------------- - 20 rolling episodes, 50 calendars per episode. - Target 20% of first-year action hours. - Mean actual deferred fraction: **21.3%**. - Mean first-year risk loss: **0.0413**. - Mean residual risk: **0.1546** (from a lower full-plan risk). - Median total cost: **47,226 CAD**, budget-feasible in 100% of scenarios. .. list-table:: :header-rows: 1 * - Case - Scenario completion - Mean risk loss - Residual risk - Budget feasible * - Risk-ranked 20% + 75-day - 47.1% - 0.0413 - 0.1546 - 100% * - Risk-ranked 20% + 105-day - 87.4% - 0.0413 - 0.1546 - 100% For context, a proportional 20% reduction has an estimated mean first-year risk loss of about 0.0448 under the same action-benefit accounting. The risk-ranked choice loses less first-year risk reduction, but its whole-action selection also has lower scheduling completion than the proportional 20% + 105-day case. Interpretation -------------- The risk-ranked 105-day case completes less of the assumed calendar than the proportional case, but it protects more first-year risk reduction per deferred hour. Neither result is a multi-year risk recommendation. A real deferral decision would need year-specific risk accounting, recurring treatment effects, and asset-owner priorities. Limits ------ - Ranking uses first-year risk reduction per action hour only. - Whole segment actions are deferred; actual deferred fraction can overshoot the target. - No multi-year risk, treatment recurrence, outage, or safety model. - Weather and outage assumptions are prototype values. - The MILP is exact only for the declared linear risk state at 59 segments.