Bayesian estimate of the elastic modulus of concrete box girders from dynamic identification: a statistical framework for the A24 motorway in Italy

Abstract

This paper delivers a reliability-based method for the assessment of the elastic modulus (EM) of concrete in simply supported girders from dynamic identification. The correlation between the natural frequencies of the first bending modes and the concrete EM supports the use of the first natural frequency as a predictor of the EM value, which is a well-acknowledged indicator of the state of concrete. In the current application, the EMs of seven girders provide the prior state of knowledge about the considered bridge class, possibly to be obtained by more samples in working applications. The identified natural frequencies update the prior probability distribution of the EMs using Bayes inference. The resulting probability of exceeding a specific EM value expresses the degree of belief of the inspector in the obtained EM. The posterior probability, compared to a proper threshold, could be used in decision-making processes when prioritising the interventions in the maintenance plans.