Actionable condition information and resource priorities in flood-disrupted urban rail recovery

Condition information supports flood-disrupted metro recovery by admitting work or confirming passenger-service paths. Its value depends on whether capacity becomes available while passengers still need service. We link condition beliefs, crew-task completion and depleting passenger stock through a common transition model, with evacuation, bus-bridging and local-operation capabilities assigned to confirmed service islands. Paired experiments use the Zhengzhou Metro Line 5 graph with synthetic damage and explicit reporting likelihoods. Under capped operation and base short-duration tasks, three-step rollout reduces mean stock loss by 7.0–11.3% relative to entropy–passenger greedy. With tripled repair durations, screened full-horizon evaluation reduces loss by 7.11–7.15% relative to three-step evaluation under common computational caps; its difference from uniform full-horizon allocation remains inconclusive. Across the tested capacity and duration grid, sparse reporting favours an additional inspection crew, whereas abundant reporting favours repair. This conditional resource priority links information availability to the work and service that crews can deliver. Mechanism and exact-network checks delimit the findings, including sensitivity to admission and clearance and cases favouring shorter evaluation. The framework supports controlled recovery-planning comparisons; field deployment and flooded-tunnel restoration forecasts require calibrated work packages, verified service capacities and independent event validation.