PSI - Issue 78

Ataklti Gebrehiwet Gebrekidan et al. / Procedia Structural Integrity 78 (2026) 1665–1672

1667

The safety checks of the bridge were made according to the current Italian regulations NTC (2018). The seismic assessment was conducted through Response Spectrum Analysis (RSA) using a 3D finite element model developed with SAP2000 (Computers and Structures, 2023). The RSA results revealed that the piers would not be able to withstand the bending moments generated by the SLV design earthquake therefore, the bridge required retrofitting to meet code seismic requirements. To improve the performance of the as-built bridge (As-B), four retrofitting strategies were considered. Three of those aim to enhance the strength, stiffness and/or ductility of the piers: Reinforced Concrete Jacketing (RC-J) and Steel Jacketing (S-J) are designed to improve structural performance by securing elastic behaviour under design-level seismic loading. FRP Confinement (FRP-C) targets improved ductility and shear strength by providing lateral confinement. The fourth intervention, Seismic Isolation (SI), aims to reduce the seismic demand by decoupling the superstructure from the substructure, enabling the piers to respond within the elastic range. Each of the four interventions has been designed for both the low and high seismic hazard sites.

Figure 1: Frontal view of the type of case-study bridge.

2.2. Detailed numerical modelling A 3D numerical model of the bridge was developed using the open-source finite element (FE) software OpenSeesPy (Zhu et al., 2018) to perform multiple nonlinear seismic static and dynamic analyses. The model Figure 2 explicitly represents both the superstructure and substructure; the piers were modelled with fully fixed supports, and the bridge-abutment and soil-structure interactions were neglected. The deck main girders are modelled using elastic beam-column elements since they are expected to remain linear elastic during the seismic excitation. The pier elements were modelled using displacement-based fibre beam-column elements, and two Gauss-Legendre integration points. P-Delta coordinate transformation was considered to account for potential second-order effects. The unconfined and confined concrete were modelled using the uniaxialMaterial Concrete02 material and the effect of confinement was accounted for, while the longitudinal reinforcement was modelled using the uniaxialMaterial Steel02 material. The bearings and the connections at the Gerber saddles were modelled using elastic equivalent springs (links), while the slab was not explicitly modelled, assigning, instead, rigid diaphragms to the deck nodes. For seismic isolators, zero length element incorporating constitutive models for their lateral, vertical, and rotational responses, were used. The lateral response is defined using a bilinear constitutive relationship, while the vertical and rotational responses are modelled with a uniaxial elastic material, as defined in Pinto et al. (2024). A 5% Rayleigh damping ratio was applied for the fundamental modes. Dead and superimposed loads, along with seismic loading, were considered. 3. Seismic hazard analysis and ground motion selection The seismic hazard at the chosen locations was defined using probabilistic seismic hazard analysis (PSHA) with OpenQuake (Silva et al., 2014). The hazard curves for both locations are shown in Figure 5. Nine intensity measure (IM) levels were considered, and a suite of 25 hazard-consistent ground motion pairings was selected for each IM level. The EzGM tool was used to scale the records chosen from the NGA-West2 database to fit the statis\tical distribution of the conditional spectrum (Ozsarac et al., 2023). The chosen IM was AvgSA, defined as the geometric mean of the pseudo-spectral acceleration in a structure-specific range of periods, depending on the first-mode period. The considered period range follows the recommendations in NTC 2018 for isolated and non-isolated structures and