PSI - Issue 84

Available online at www.sciencedirect.com

ScienceDirect

Procedia Structural Integrity 84 (2026) 392–400

© 2026 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the Conference Keywords: Existing bridges; Fragility assessment; Traffic loads; Corrosion; Machine Learning. Abstract Italian and European existing bridges are approaching the end of their service life and, to ensure required capacity to withstand exogenous hazards, timely monitoring and appropriate intervention policies are required. The risk classification proposed by the Italian guidelines involves time- and cost-consuming activities for large scale prioritization and alternative approaches should be explored. This paper proposes a Machine learning (ML)-based framework for fragility assessment of multi-span simply supported prestressed reinforced concrete bridges under traffic loads. A new taxonomy was used to this scope and finite element (FE) models were used for parametric analyses accounting for material degradation (i.e., corrosion). The results were used to train a ML classification algorithm (i.e., eXtreme Gradient Boosting, XGBoost) to derive fragility curves at collapse limit state for flexural and shear failure mechanisms. Statistical metrics were used to quantify XGBoost performance as ML surrogate model an eXplainability approach was used to identify the most critical input features. Fragility curve derived by XGBoost predictions and numerical analyses were compared for validation. The proposed ML surrogate model is not a surrogate for numerical analysis, i.e., the single output cannot be mistaken for the result of a numerical analysis, but overall, the set of predictions can be used to derive fragility curves within the input parameter space. The main innovation is a tangible and quantitative set of ML-taxonomy-based fragility curves as support for prioritization of bridge portfolios. III Fabre Conference: Existing Bridges, Viaducts, and Tunnels: Research, Innovation, and Applications ML-based framework for fragility assessment of existing reinforced concrete bridges under traffic loads and material degradation Mirko Calò a, *, Alessandro Nettis a , Andrea Nettis a , Sergio Ruggieri a , Andrea Dall'Asta b , Giuseppina Uva a a DICATECH Department, Polytechnic University of Bari, Bari, Italy b School of Science and Technology, University of Camerino, Camerino, Italy

* Corresponding author. E-mail address: mirko.calo@poliba.it

2452-3216 © 2026 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the Conference 10.1016/j.prostr.2026.06.051