Stochastic-based framework for comparing risk estimates of multi-span masonry arch bridges subject to seismic and flood hazards

• Proposal of a performance indicator based on the ratio of vulnerabilities associated with two different assets. • Inclusion of stochastic surrogate models for the derivation of non-log-normal fragility surfaces. • Seismic assessment of the out-of-plane behavior of slender multi-span MABs considering flood-related hazard uncertainties Railway networks are highly centralized and operate as series systems, where failure of a single component, e.g. , bridges, can disrupt the entire line. Masonry arch bridges (MABs) account for approximately 40% of the bridge stock in many European railway networks and often exceed their design operational life. Their vulnerability can increase significantly when exposed to natural hazards, particularly for larger and slender multi-span configurations susceptible to global out-of-plane failure mode. Therefore, quantifying and comparing the multi-hazard vulnerability of these assets is essential to identify critical infrastructure. This research proposes a risk-based framework for assessing out-of-plane failure of multi-span MABs under seismic and flood hazards. The framework employs stochastic surrogate models to derive fragility surfaces directly in the hazard domain without assuming log-normality. Joint hazard probabilities are used to account for asset exposure, considering peak ground acceleration and flow discharge as intensity measures for seismic and flood hazards, respectively. The resulting risk curves support the prioritization of infrastructure assets across hazard-intensity levels. The framework is demonstrated by comparing monetary risk estimates for two multi-span MABs identified as critical infrastructure in the Portuguese railway network.