Modeling building-aware overland flow and dynamic collapse during hurricane surge

Traditional storm surge models represent urban areas using parameterized surface roughness, which captures bulk resistance effects but does not resolve building-scale flow–structure interactions or their influence on neighborhood-scale hydrodynamics. Laboratory and field evidence shows that structures redirect currents, accelerate flow through gaps, and create sheltered wake zones–affecting hazard intensity at neighborhood scales. Here, we present a building-aware surge model with dynamic collapse capability, applied to Estero Island, Florida, during Hurricane Ian. Using Delft3D-FM with real-time control, we simulate how building presence and failure reshape local hydrodynamics. While surge heights remain similar across scenarios, building inclusion alters flow velocity and shear patterns, producing local velocity increases up to 1.0 m/s near corridors and amplifying shear near corners and constrictions. The collapse of structures triggers nonlinear feedbacks, rerouting flow and exposing sheltered areas, and increasing local velocities by 40–60% within 25–50 m of the collapse location, transitioning from 1.0 m/s to 1.4–1.6 m/s. This result underscores the nonlinear amplification of flow in response to structural failure. • Introduces a building-aware storm surge modeling framework using Delft3D-FM with explicit hydraulic-wall representations of urban structures. • Implements a real-time control (RTC) approach to simulate dynamic building collapse and evaluate its feedback on overland flow pathways. • Demonstrates that buildings, while not altering basin-scale surge heights, modulate local velocity fields, creating gap jets, wake zones, and shear hotspots. • Shows that collapse of a single seaward structure can re-route currents, amplify local velocities, and generate new downstream scour-prone areas. • Provides a scalable, physics-based methodology for integrating structural evolution and damage-state information into urban coastal hazard models.