PSI - Issue 84

Stefano Bozza et al. / Procedia Structural Integrity 84 (2026) 686–693

688

3. Analyses 3.1. FE model The FE model of the bridge was assembled in SAP2000 v25.2.0 (CSI 2016), using the software API and Python programming to import nodes coordinates and cross sections of girder and piers. The box girder was modelled via elastic frame elements, assigning a single element to each precast segment, using both constant and tapered sections according to the geometry of the bridge. Moreover, elastic frame elements were also used to reproduce the actual cross section of the shaft of the piers, while rigid elastic elements were assigned to the pier cap. Non linearities were concentrated in the load bearings, in the shear keys, at the base of the piers and in the prestressing cables. Load bearings devices were modelled as non-linear link elements, fixing the vertical degree of freedom, assigning a very small stiffness along the rotational degrees of freedom, and an initial high stiffness plus a plastic behaviour to the horizontal degrees of freedom. For free horizontal degrees of freedom, the plastic branch was set at a small force (approximatively 3% of the vertical reaction due to dead loads) in order to model in a simplified way friction, while for restrained degree of freedom the plastic branch was set equal to the nominal strength of the device. Furthermore, the shear keys were also modelled with elastic-plastic link elements, assigning a nonlinear behaviour along the only restrained degree of freedom and an elastic behaviour with negligible stiffness along other directions. On the other hand, fiber-based non-linear hinges were placed at the bottom of each pier, considering a hinge length equal to 10% of the pier height. An elastic-plastic behaviour with kinematic hysteresis was assigned to each steel fiber representing the rebars, while Mander constitutive law and Takeda hysteresis were used to define the non-linear cyclic behaviour of concrete fibers. Each pier section was discretized into 92 steel fibers and 158 concrete fibers. Finally, prestressing tendons in the piers were modelled as elements with an elastic-plastic behaviour. Furthermore, the bridge has deep foundations, thus soil-structure interaction was neglected, using fixed restraints at the base of each pier. Moreover, since abutments are very stiff elements rigidly connected to the ground, they were not explicitly included in the FE model, thus assuming at the top of the abutments the same motion of the ground. 3.2. Nonlinear time-history analyses The seismic performance of the bridge was assessed via nonlinear dynamic time-history analyses. A nominal life of 100 years was taken into account, as suggested by the Eurocode 0 for bridges (EN 1990 (2002)), for a return period of 949 years, corresponding to the life safety (LS) limit state probability of occurrence. Seven triplets of accelerograms were chosen from the Engineering Strong Motion Database (Luzi et al. 2020) using the REXELweb tool (Sgobba et al. 2019), assuming soil category B and flat topography. Moreover, the signals where selected to match the Life Safety (LS) response spectrum according to NTC (2018). The accelerogram records used for the nonlinear time-history analyses are reported in Table 1.

Table 1. Accelerogram records used in the time-history analyses.

No.

Event

Network Station

Mw Rep (km)

PGA (cm/s 2 )

1 EMSC-20140203_0000008

HI

ARG2

6.0

9.6

257 284 163 211 295 220 194

2 GR-1981-0001

HL

XLCA 6.6

35.9 11.3 13.7 11.2 76.3 40.5

3 EMSC-20161026_0000077 4 EMSC-20161026_0000095 5 EMSC-20161026_0000077 6 INT-20230206_0000222 7 INT-20230206_0000222

IT IT IT

NOR NOR NRC

5.5 5.9 5.5 7.5 7.5

TK 3802 TK 4611