PSI - Issue 78

Carmine Lupo et al. / Procedia Structural Integrity 78 (2026) 185–192

190

particular, to keep the model computationally efficient , a “Spine” configuration was selected. The piers and deck were modelled using frame elements positioned along the centroid axes of the actual components. Various types of link elements were employed to simulate the behaviour of the bearings as well as the plastic hinges at the bases of the piers. The model also incorporated appropriate joint elements to represent the structural scheme, along with necessary constraints. A total of 300 nonlinear dynamic analyses were conducted, managed via a MATLAB script. The fragility functions were evaluated through a Multiple Stripe Analysis (MSA). In addition to the fragility curves developed for individual EDPs, overall system fragility curves were also defined. According to this approach, the probability of system failure is bounded between a lower and an upper limit. The lower bound corresponds to the maximum failure probability among the individual EDPs, while the upper bound represents a combined probability derived from the fragility curves of all considered EDPs, see Petti and Mammone (2019). This method is particularly applicable to series systems, in which the failure of a single component results in the failure of the entire system (Melchers, 1999). In order to analyse the structural response and to evaluate fragility curves, it was necessary to define a set of accelerometric records representative of various seismic events. To this end, 100 accelerograms were selected corresponding to real earthquakes that occurred over the years in different parts of the world, exhibiting various characteristics but falling within sufficiently broad ranges of the aforementioned intensity measures (IMs). This method avoids the need to scale individual accelerograms, a process that may introduce unrealistic seismic demands, and reduces site-specific dependencies, thereby improving the general applicability of the analysis. Among the 100 accelerometric records, two broad categories can be identified: ‘Near Fault’ and ‘Far Field’ ground motions (Billah et al, 2013). The main distinction between these categories lies in the epicentral distance. Specifically, the ‘Near Fault’ category includes 60 records with epicentre distances between 0 and 15 km, while the ‘Far Field’ cat egory includes 40 records with epicentre distances greater than 15 km. Table 1 provides a summary some of the selected records. The recordings were taken from the following online sources: CESMD Strong-Motion Data-Set, NCEI Earthquake Strong-Motion Database, PEER Ground Motion Database, ESD the European Strong-Motion Database, University of Chile (Ruben Boroschek, Pedro Soto, Ricardo Leon). 3.3. Seismic Demand

Table 1. Seismic Demand – Earthquakes Signals and related IMs n. Event Station

Epicentre Distance [km]

Magnitude PGA [g]

Ia [m/s]

Sabetta V

1 2

Chi-Chi Chi-Chi

CHY 080

0.2 0.7

7.6 7.6

0.82 0.36

9.65 2.97

18.19 18.06

Tcu052

:

:

:

:

:

:

:

:

99

Amberley (NZ)

Kekerengu Valley road Building Research Institute

115.6

7.8 7.5

1.04

10.39

4.42 3.34

100 Bucharest (RO)

161

0.2

0.80

4. Results The analysis of the results involved evaluating the correlation between the considered EDPs and IMs. The objective was to identify which of the aforementioned intensity measures minimised the dispersion of the results. To this end, Figure 3.a presents the results obtained for the plastic hinge rotation at the base of the Short Piers and the Sabetta V IM. A quadratic function was found to provide the best fit across all EDPs, and using a logarithmic scale yielded higher correlation values. Figure 3.b shows a comparison of the best-fitting functions corresponding to the three retrofitting strategies for the same EDP. In Figure 3.c, an example of fragility curves is presented for the plastic hinge rotation at the base of the Short Piers.