PSI - Issue 78

Marco Civera et al. / Procedia Structural Integrity 78 (2026) 1783–1790

1784

Keywords: resilience; road network; network-level assessment; seismic damage; damage state.

1. Introduction This research work is a direct continuation of the analyses reported in (Miano et al. 2024). As better detailed there, the intention is to evaluate the resilience of road networks (RNs) in the aftermath of a seismic event. This is in line with the most recent advances in infrastructure maintenance, according to the Italian regulations, which prescribe assessment at the network level for critical infrastructures ranked high in multi-risk assessments – please refer to the definition of Level 5 analysis in the recently adopted Linee guida per la classificazione e gestione del rischio, la valutazione della sicurezza ed il monitoraggio dei ponti esistenti (Consiglio Superiore dei Lavori Pubblici 2020). Also, these analyses are particularly needed to establish contingency plans for Civil Protection and ensure disaster preparedness. In fact, RNs play a pivotal role during emergencies, facilitating evacuation, relief supply, and interconnectivity between critical nodes such as hospitals and emergency facilities. However, the operability of these networks is often compromised by seismic events, which can lead to severe disruptions due to structural failures – both direct hits (bridge failures) or indirect consequences (surrounding buildings collapsing and obstructing the road paths). Such interruptions impact immediate disaster response, as was investigated in (Miano et al. 2024), but also have long-term consequences for mid-term socio-economic recovery. This study focuses on the latter aspect. 2. Previous Methodology: Post-Event Efficiency Assessment To date, several gaps remain in understanding the broader system-level performance of urban road networks during seismic events. Indeed, the interconnectedness of transportation systems, combined with the cascading effects of surrounding structure failures, necessitates a multidisciplinary and holistic approach to assessing network resilience. To this aim, the methodology outlined in (Miano et al. 2024), and followed here as well, consider a comprehensive evaluation of both network components and adjacent structures. In particular, seismic demand and capacity, expressed as peak ground acceleration (PGA), are derived from probabilistic seismic hazard analysis (PSHA). Site-specific PGA values are assigned to each RN component using established ground motion prediction equations (GMPEs). The vulnerability of the RN components, i.e. bridges and adjacent buildings, is assessed using fragility curves (FCs). These curves estimate the probability of reaching specific damage states based on PGA. The concept was, therefore, that damage states exceeding moderate damage trigger road closures, while a partial or complete collapse of nearby buildings will block the adjacent roads. Regarding buildings, for simplicity, they have all been assumed to be constructed with reinforced concrete (RC), following the FCs proposed by (Rosti et al. 2021) and the five damage levels considered prescribed by EMS 98 (Grünthal and Schwarz 1998). For the road bridges, instead, the distributions developed by (Moschonas et al. 2009) were considered.The concept is quite straightforward: each structure (bridge or building in the RN) is assigned a logic value of 1 (road disruption) if it meets the condition > (1) Where = 50 , = 50 (2) are, in this order, the seismic demand (derived from the INGV grid map of the area) and the seismic capacity, obtained via the FCs mentioned above. In both equations, is a uniformly distributed pseudorandom scalar, while 50 and 50 are the 50th percentile of the lognormal distribution of demand and capacity, respectively, and β andB denote their respective standard deviations. If the criterion is not respected – i.e. the seismic demand is lower than the capacity of the target asset, a logic default value of 0 (no disruption) is kept. Hence, the logic is binary, and one building or bridge can only be considered as disrupting or not the RN, independently of the level of the disruption and, therefore, of the time required to reactivate it. This aspect will be recalled later for further considerations. As mentioned, unlike traditional assessments, the proposed methodology integrates probabilistic seismic hazard