Agent-based Vulnerability Model for Pedestrians Exposed to Urban Floodwaters

Abstract

Climate change-induced floods will have a profound impact on densely populated urban areas. The survey results indicate that a substantial proportion of respondents engaged in evacuation behavior during urban flooding events. However, current assessment methods may underestimate the impact of human motions in floodwaters on pedestrian evacuation safety. To quantitively study the dynamic vulnerability of individuals exposed to flooding scenarios, an agent-based vulnerability model was proposed based on mechanics modelling and experimentally calibrating. A full-scale physical testing platform was constructed and utilized to calibrate the proposed model and to determine the stability limits of pedestrian safety in floodwaters. Spatial and temporal dynamic characteristics of pedestrians were analyzed and results reveal significant variations in pedestrian movement and stability. The general temporal trend of movement speed changing as a power function of the specific flood force has been validated. It is also found that pedestrian stability is notably affected by movement in floodwaters, particularly when walking against the flow, which intensifies the risk of instability, leading to vulnerability indices that increase by 123.2% at a depth of 0.3 m and by 82.7% at 0.5 m compared to still-water conditions. In contrast, moving with the flow reduces hydrodynamic forces, although the rate of this reduction decreases with greater water depths, dropping to 16.0% at 0.5 m and 9.7% at 0.7 m. Additionally, this work provides guidelines for assessing pedestrian evacuation vulnerability that enhances evacuation safety and supports flood management.