Probabilistic framework for seismic performance assessment of a multi-span masonry arch bridge employing surrogate modeling techniques

• Published in: Engineering structures Vol. 325; p. 119399
• Main Authors: Cabanzo, Carlos, Mendes, Nuno, Akiyama, Mitsuyoshi, Lourenço, Paulo B., Matos, Jose C.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 15.02.2025
• Subjects: Adaptive sequential sampling; Non-destructive testing; Seismic fragility; Structural parameter uncertainty; Support vector machine for classification; Adaptive sequential sampling; Non-destructive testing; Seismic fragility; Support vector machine for classification; Structural parameter uncertainty
• ISSN: 0141-0296; 1873-7323
• DOI: 10.1016/j.engstruct.2024.119399

Extreme natural events, such as earthquakes, are unavoidable and have been identified as triggers of bridge failure. In Portugal, MABs with larger spans and thinner piers were built in the later years of the railway expansion and are expected to continue operating as a crucial part of the railway network without elevated maintenance costs. Large MABs have been the subject of research related to probabilistic analyses based on experimental information, and their elevated computational cost has been identified as one of the main challenges. Moreover, there is limited research regarding the seismic performance of large MABs and the effects of uncertainties in their out-of-plane behavior. The present research aims to provide a framework for assessing the seismic performance of large MABs, based on information from non-destructive tests, through the determination of the structural reliability by implementing surrogate models based on material uncertainties. To ensure the accuracy of the reliability analysis, adaptive sequential sampling was employed to train the surrogate model near the performance limit states. The framework was applied to assess the structural reliability of the Quinta Nova bridge, resulting in the definition of seismic fragility curves based on the Eurocode performance limit states and peak ground acceleration for the seismic action. [Display omitted] •Application of support vector machine for structural fragility.•Integration adaptive sequential sampling for surrogate model optimization.•Structural fragility based on experimental data and parameter uncertainties.•Code-based seismic performance assessment to aid asset management.