Data-driven risk assessment of climate-related failures in railway infrastructure

Climate-related failures pose increasing risks to railway operation and maintenance, requiring robust assessments to support effective resilience planning. Accordingly, this study presents a comprehensive risk assessment to identify, quantify, and prioritize climate-related failure modes (CRFMs). CRFMs were identified through text mining of 15-year corrective maintenance records across all five Swedish railway regions. A probabilistic model was then developed to analyze the risk of identified CRFMs, considering both operational and maintenance costs. Risk distributions were derived using Monte Carlo simulation and summarized by Expected Value of Risk (EVoR) and Conditional Value at Risk (CVaR 0.90 ). As a result, 14 CRFMs were identified, accounting for 47% of all failure records. Their annual trend and seasonal distribution align well with periods of increased extreme weather events and dominant seasonal climate hazards. Furthermore, clustering the regional profile of CRFMs reveals regional similarity. The CRFMs were then prioritized, showing that 4 modes (i.e., Track deformation, Snow and ice, Buckling, and Rail breakage/crack) account for 73% of total CVaR 0.90 . The amplification of total risk in the tail indicates a 74% higher cost burden under extreme conditions. Finally, regional climate vulnerability was assessed using risk metrics normalized by track length and traffic density.