Abstract
Flooding presents a growing threat to urban road networks, compromising their structural integrity and reducing their functionality during extreme weather events. With the increasing frequency and severity of floods due to climate change, there is a critical need for comprehensive methods to assess the vulnerability of road networks and manage traffic resilience under such conditions.Existing research often simplifies flood impacts through binary classifications of road usability and overlooks the combined influence of flood depth and flow velocity on different vehicle types. Moreover, most studies neglect dynamic traffic behaviours, such as congestion propagation and vehicle rerouting, limiting the practical applicability of risk assessments in real-world flood scenarios. This thesis addresses these research gaps by developing an integrated, multi-stage framework that evaluates road network functionality loss and traffic performance under both deterministic and probabilistic flood conditions.
The methodology comprises three stages. First, a deterministic approach overlays historical flood hazard maps onto a graph-based road network topology, using vehicle-specific roadworthiness criteria derived from flood stability experiments to assess severity. Graph metrics—such as average node degree, clustering coefficient, and shortest path—are used to evaluate road network functionality loss, and time-based isochrones assess hospital accessibility under different vehicle types and flood intensities. Second, a probabilistic framework is introduced to account for uncertainty in vehicle stability. Flood hazard curves (based on return periods) are combined with fragility curves (representing the probability of vehicle instability) to derive the likelihood of road functionality loss. A hazard-agnostic loss probability is calculated through weighted convolution, enabling predictions independent of any single flood scenario. Third, an agent-based transport simulation (MATSim) models dynamic traffic flow in the flood-affected road network. It simulates vehicle redistribution, congestion evolution, and changes in accessibility pre- and post-flood, using free-speed reductions derived from probabilistic functionality loss as input parameters.
Key findings reveal that vehicle vulnerability significantly varies with flood characteristics, affecting both individual road accessibility and network-wide performance. Roads with moderate flood exposure may remain structurally intact but functionally impaired due to reduced speeds or instability risks, particularly for lighter vehicles. The hazard-agnostic probabilistic framework allows practitioners to estimate functionality loss without relying on future flood event data, making it applicable in data-scarce regions. Agent-based simulations show that network disruptions cause substantial traffic redistribution, with specific roads emerging as congestion hotspots, and hospital accessibility shrinking significantly during flood-induced congestion.
These findings have direct implications for flood risk management, transport planning, and emergency response strategies. The deterministic and probabilistic frameworks provide planners with tools to map and quantify vulnerability across a network, while the dynamic traffic model supports real-time decision-making, such as prioritising road repair, planning vehicle detours, and identifying critical links for resilience upgrades. The integrated methodology offers a robust basis for enhancing transport network preparedness in the face of increasing flood risks.
| Date of Award | 30 Sept 2025 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Raffaele De Risi (Supervisor) & Neil J Carhart (Supervisor) |