Fulvio Parisi

Most of the Italian infrastructures have been built since the mid‐twentieth century. Prestressed concrete (PC) has been widely used, especially for bridges and viaducts. These structures have proved effective over time, but their performance could be affected by degradation phenomena (i.e., corrosion of the tendons and residual prestressing) or construction defects (i.e., grouting of the ducts), which need to be accurately modeled. This paper focuses on the flexural response of two reduced‐scale, posttensioned, PC bridge girders, prestressed with two different jacking forces and tested under four‐point bending configuration. Four multidimensional numerical models were developed to simulate the experimental behaviors of the selected specimens, using two alternative computational strategies, namely, the finite element method (FEM) and applied element method (AEM). Three FEM modeling were considered, ranging from one‐dimensional lumped‐plasticity models to two‐dimensional and three‐dimensional (3D) spread‐plasticity models. Besides, 3D AEM models were developed into another software package. The accuracy and the computational costs of the two numerical strategies are discussed in this paper. Also, the main features of each numerical modeling technique are addressed, including comparisons between numerical and experimental global response data as well as the statistical characterization of the model error. Based on these different features, the authors also suggest the most suitable numerical strategy for future studies on PC girders.

After the 2018 collapse of Morandi highway bridge in Genova, new Italian guidelines (GL) were issued for prioritization of safety evaluations and retrofit interventions on existing bridges. Among several novelties, GL introduced new traffic load models (TLMs) which can be adopted for existing bridges in case of noncompliant safety checks according to the Italian building code (NTC) provisions. In this study, a class of simply supported, beam‐type, prestressed concrete bridge decks built between 1970 and 1980 in Italy is considered. Assuming geometric, material, and load random variables together with consideration of capacity model uncertainty, Monte Carlo sampling technique was implemented in MATLAB in order to randomly generate deck models and to evaluate their traffic‐load fragility. Fragility analysis was carried out to compare vulnerability levels of existing Italian bridges under different load patterns provided by GL and NTC. In addition to a prescribed TLM, the sensitivity of fragility to bridge usage limitations, such as reduced distance of external load lane from kerb or reduced number of lanes, was also quantified to support decision‐making by road management companies. In the final part of the study, the annual failure probability of selected bridges was estimated using a European weigh‐in‐motion database and convolution of fragility and hazard. Analysis results show that structural fragility significantly depends on the load pattern, indicating that more realistic vehicle models should be developed to achieve a target safety level conforming to modern codes for constructions.