PSI - Issue 78

Andrea Nettis et al. / Procedia Structural Integrity 78 (2026) 1412–1419

1418

seismic demand overestimation induced by the simplified response analysis is observed. These inaccuracies lead to overestimations of the damage probability and of the loss curves for low value of the IM in fragility curves.

Fig. 3. Expected Annual Losses computed by CSM and NLTHA within multi-stripe analysis.

Fig. 4. Seismic demand data, fragility (DS1 and DS4) and vulnerability curves for the bridge B222 analysed in the longitudinal (a-b-c) and transverse direction (d-e-f). 5. Remarks and future developments As a first step towards the practical implementation of loss-based design for multi-span bridges, this study evaluates the accuracy of simplified loss assessment strategies. A set of 18 case-study continuous-deck bridges with single column piers, located in a high-seismicity region, is analysed. For each bridge, benchmark estimates of the expected annual loss (EAL) are obtained through nonlinear time-history analyses (NLTHA). These benchmark values are compared against estimates derived from a simplified alternative method deriving the force-displacement curve analytically with a displacement-based approach and deriving demand using the capacity spectrum method (CSM) with as-recorded ground-motion spectra. Fragility analyses are conducted using both the cloud method — employing a power-law functional form — and the multi-stripe method. The CSM appears promising for future non-iterative loss based design applications. When combined with a cloud approach, it produces loss estimates with satisfying accuracy in the analysis in longitudinal direction, providing slight overestimations compared to NLTHA-based expected annual losses. In contrast, using a multi-stripe approach, the relative error on average loss increases up to 40%. However, these discrepancies remain lower than the ones derived using cloud analysis instead of the multi-stripe approach, both based on NLTHA. The errors of the CSM increase significantly in the transverse direction. However, these could be less critical from a design standpoint based because the losses in the transverse direction are lower than those calculated