Reliability assessment of evacuation systems at the urban spatial scale using a two-stage evacuation simulation under temporary and fixed refuge scenarios

The reliability of emergency evacuation systems is crucial for urban resilience and public safety, directly impacting disaster response effectiveness and system stability. While evacuation in severe disasters typically involves multiple stages, the dynamic performance of such staged systems at a large-scale requires further exploration. This study proposes a two-stage evacuation simulation model to analyze the status of demand points and shelters under temporary and fixed refuge scenarios, identifying areas of insufficient coverage and assessing overall system performance. The innovations lie in simulating evacuation at the urban spatial level and considering the staged nature of evacuations in post-disaster personnel transfer dynamics. The main findings include: (i) evacuation efficiency is generally higher in the temporary refuge stage, characterized by rapid initial response but localized saturation risks; (ii) shelter distribution and residual capacity critically determine system performance, with well-distributed shelter networks exhibiting higher reliability; (iii) the number of available shelters around demand points influences outcomes, promoting load distribution and redundancy in the first stage but revealing capacity bottlenecks in the second. This study delivers a transferable simulation framework and empirical evidence on the trade-off between stage efficiency and stability, demonstrating that spatial equity in shelter distribution is foundational to urban evacuation reliability.