Time-Window KPIs for Evaluating Urban Flood Evacuation Simulations: Definitions, Computation, and a Case Demonstration

Evaluating flood evacuation performance using a single “overall success” indicator can be misleading, because it does not distinguish evacuees who reach safety within an operationally critical time window from those who arrive late. This short paper formalizes a time-window evaluation set for urban flood evacuation simulations and provides an implementation-oriented demonstration using a real-world case study. We define Success Rate as the proportion of agents that reach a shelter during the simulation horizon, and introduce two complementary time-window metrics at a critical time threshold T = ۶۰ min: KPI_CORE, the share of the total population that has not reached a shelter by time T (regardless of eventual arrival), and KPI_DELAY, the share of the total population that reaches a shelter only after time T (t ≥ T). To capture evacuation tempo beyond endpoint ratios, we also report clearance-time landmarks T۵۰/T۸۰/T۹۰, defined as the times required for ۵۰%, ۸۰%, and ۹۰% of evacuees to reach shelters. All indicators are computed directly from agent-level logs and aggregated into a post-simulation table for reporting. The proposed KPI set enables consistent reporting across scenarios, highlights the “tail risk” of late arrivals, and supports operational decision-making where time constraints dominate. A brief case demonstration illustrates how the combined indicators provide a clearer picture than success rate alone.

کلیدواژه ها:

Flood evacuation ، Performance indicators ، Time-window KPI ، Evacuation curve ، Agent-based simulation ، Urban resilience