PSI - Issue 84

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

524

studies by Du et al. (2005) and Opabola (2022), consistent with the advanced deterioration models discussed in Di Mucci et al. (2025b). Corrosion of the transverse reinforcement also impairs confinement effectiveness, reducing strength and ductility of confined concrete. This mechanism was captured using a corrosion-adjusted Mander formulation (Mander et al., 1988). Damage was assigned independently on each face of the pier to account for spatial variability of corrosions and CV-informed inspection data. 3.2. Time-dependent corrosion evolution and state-based formulation To overcome the high aleatory uncertainty associated with the corrosion initiation time ( T i ) (Bertolini et al., 2013; Choe et al., 2008; Engelund et al., 2000), a state-based degradation strategy was adopted, where the observed mass loss ( Q corr,obs ) at inspection time (Present – t 0 ) is used as the primary state variable. Instead of estimating T i from diffusion parameters, the methodology restricts modelling to the propagation phase of corrosion, using the piecewise formulation by Cui et al. (2018), where the corrosion rate λ(t) governs the evolution of the corrosion state through its cumulative effect on the residual reinforcement area: ( )= 0 − ( ) 0 ; (1) ( )= 4 [ 0 − 2 ∫ ( ) ] 2 (2) where A 0 and d s0 denote the pristine cross-sectional area and diameter of the steel bars, respectively, while A r (t) is the residual bar cross-sectional area at time t (Section 3.3). The corrosion evolution is described through the following phases: 1. Phase 1 (Initiation to cracking): Triggered when T i t spalling , the reinforcement is directly exposed to the environment. In this third phase, where BriCANet can be applied, corrosion severity is mapped to CV-predicted intensity classes, which are associated with the previously defined mass loss intervals. For pre-spalling conditions (Phases 1–2), where BriCANet cannot be applied, the current state ( t 0 ) is inferred through CV-based defect detection (Ruggieri et al., 2025). Cracked faces are classified as Phase 2 with Q corr ∈ [5,10]% while visually intact faces are classified as Phase 1 with Q corr ∈ [0,5]% consistent with CONTECVET. 3.3. Numerical implementation and forecasting The adopted modelling strategy is based on the individual pier model which can be efficiently used for simply supported girder bridges (Nettis et al., 2024). For redundant bridge structural schemes, such as continuous-deck bridges, more complex numerical models should be used. The seismic response of a corroded bridge pier was simulated using a fiber-based multi-degree-of-freedom (MDoF) model in OpenSees. The pier was discretized into two force-based beam–column elements: a lower segment (10% of the total height) representing the plastic hinge region— where corrosion effects are localized—and an upper pristine segment. Further information on the modelling strategy is available in Di Mucci et al. (2025a). The modeling approach treats the four faces of the pier independently to account for spatial variability of corrosion and CV-informed inspection data. The fiber section of the lower element was partitioned to assign differentiated Q corr values to the longitudinal and transverse reinforcement of each face. Concrete fibers were modeled using the Concrete01 material, allowing for cover removal when Q corr > 10%, while reinforcement was simulated using the Steel02 model. Bearing devices were introduced as linear elastic springs via twoNodeLink elements.