Risk Assessment and Vulnerability Identification of Energy-Transportation Infrastructure Systems to Extreme Weather

The interaction between extreme weather events and interdependent critical infrastructure systems involves complex spatiotemporal dynamics. Multi-type emergency decisions within energy-transportation infrastructures significantly influence system performance throughout the extreme weather process. A comprehensive assessment of these factors faces challenges in model complexity, heterogeneous differences between energy and transportation systems, and cross-sector privacy. This paper proposes a risk assessment framework that integrates the heterogeneous energy and transportation systems in the form of a unified network flow model, which enables full accommodation of multiple types of energy-transportation emergency decisions while capturing the compound spatiotemporal impacts of extreme weather on both systems simultaneously. Based on this framework, a targeted method for identifying system vulnerabilities is further developed. This method employs neural network surrogates to achieve privacy protection and accelerated identification while maintaining consideration of system interdependencies. Numerical experiments demonstrate that the proposed framework and method can reveal the risk levels faced by urban infrastructure systems, identify vulnerabilities that should be prioritized for reinforcement, and strike a balance between accuracy and speed.

💡 Research Summary

The paper addresses the growing challenge of assessing and managing the risk posed by extreme weather events to interdependent energy and transportation infrastructure systems. Recognizing that traditional resilience studies often focus on a single system or on post‑disaster restoration, the authors develop a comprehensive risk‑assessment framework that spans the entire lifecycle of an extreme weather event—pre‑event, during, and post‑event—and that explicitly incorporates multiple types of emergency decisions across both domains.

The core of the methodology is a spatiotemporal‑augmented network‑flow model that unifies power, heat, and traffic networks into a single graph. Each node and edge carries time‑dependent attributes such as state‑of‑charge (SOC), road availability, and damage status, allowing the model to capture the bidirectional influence between systems (e.g., power outages reducing charging‑station availability, or flooded roads delaying repair crews). Emergency resources—including topology reconfiguration, mobile generators, electric‑vehicle charging/discharging, fuel transport, and repair crew scheduling—are modeled as decision variables that impose capacity, cost, and time‑delay constraints on the flow.

To represent the stochastic nature of extreme weather, the authors employ a parameter‑based meteorological model (e.g., Rankine‑Vortex) and generate high‑dimensional samples using Latin Hypercube Sampling (LHS). Each sample defines a specific realization of weather parameters, which is fed into the integrated network model. Monte‑Carlo simulation repeatedly evaluates the system performance under these realizations, producing probability distributions of key metrics such as unmet power load, heat shortfall, and traffic delay.

Because solving the full mixed‑integer network model for thousands of Monte‑Carlo runs is computationally prohibitive, the paper introduces neural‑network surrogate models. These surrogates are trained on a limited set of exact model evaluations and then linearized to preserve interpretability and to enable privacy‑preserving computation: sector‑specific data never leave the surrogate, avoiding direct data sharing between power, heat, and transportation operators. The surrogates achieve speed‑ups of an order of magnitude while maintaining a mean absolute percentage error (MAPE) of 3–5 %.

Vulnerability identification is performed by sensitivity analysis on the surrogate. By differentiating the surrogate outputs with respect to node/edge parameters, the authors construct vulnerability curves and rank infrastructure elements according to their contribution to risk metrics. The results highlight critical nodes such as major substations co‑located with high‑traffic charging stations and key heat‑network pipelines, which exhibit the greatest impact on system resilience under simulated storms.

A case study on a realistic city‑scale network (thousands of nodes, tens of thousands of arcs) demonstrates the framework. In a hurricane scenario, road flooding disables several charging stations, extending the restoration time of the power system by roughly 30 %. In a snowstorm scenario, road closures delay repair crew travel, causing an average two‑day delay in power‑line repairs. The vulnerability analysis guides targeted reinforcement actions, such as dual‑redundant charging facilities, pre‑designated repair routes, and coordinated dispatch of mobile energy resources.

The paper concludes that the integrated spatiotemporal network model, combined with LHS‑Monte‑Carlo risk quantification and neural‑network surrogates, provides a scalable, privacy‑aware tool for city planners and operators to assess risk, prioritize investments, and support real‑time decision making under extreme weather conditions. The approach is extensible to additional infrastructure sectors (e.g., water, communications) and to emerging emergency resources, offering a pathway toward holistic, cross‑sector resilience management.