PSI - Issue 84

Vincenzo Mario Di Mucci et al. / Procedia Structural Integrity 84 (2026) 521–528

525

Future corrosion evolution was estimated through a state-based procedure, by solving the inverse problem Q corr,obs = f(t*) to infer a "virtual exposure time" ( t * ) on the Cui et al. (2018) corrosion curve. Since the corrosion rate (i.e., slope) is independent of the absolute T i value, the future mass loss for each face at +n years was projected as Q corr,future = f(t* +n) , allowing for an estimation of the long-term structural decay. The general framework is shown in Fig. 2. 3.4. Seismic fragility and risk assessment A cloud-based fragility analysis was conducted following the methodology of Jalayer et al. (2017) and Nettis et al. (2021), using a suite of 100 natural ground motions from the SIMBAD database. The spectral acceleration at the first mode period, S a (T 1 ), was selected as the Intensity Measure (IM), while the normalized pier drift served as the Engineering Demand Parameter (EDP). Prior to fragility assessment, static pushover (SPO) analyses were performed on 300 stochastic realizations to determine the EDP thresholds for four damage states (DS) according to Cardone (2014). By integrating the fragility functions with site-specific hazard curves, the Mean Annual Frequency of Exceedance (MAFE) was computed to assess the escalation of seismic risk over a 30-year horizon.

Fig. 2. Proposed framework for time-projected risk assessment.

4. Case-study application 4.1. Description

The presented framework was applied on a demonstrative case study two-span simply supported girder bridge. The substructure of the bridge is composed of a single-column circular pier (diameter =2 m, height =8 m) exhibiting corrosion localized at the base. Face-specific damage at the pier base was assessed from CV-inspected photographs, classifying each face as either pre-spalling (Phases 1–2) or post-spalling (Phase 3). For post-spalling faces, inspection images were further analyzed using BriCANet, which predicted Medium corrosion for the North and East faces and Low corrosion for the South and West faces. Corresponding reinforcement mass-loss values ( corr ) were then sampled to define the present state ( t 0 ). Geometric dimensions were treated as deterministic, based on accurate design documentation, whereas mechanical properties and corrosion severity were modeled probabilistically to capture