In many countries concrete bridges are reaching the end of their service-life, showing signs of deterioration, e.g. due to corrosion. Hence, the question arises whether their safety level is still acceptable. To improve the estimate of the safety level, information extracted from different tests, e.g. proof-loading, operational modal analysis, etc., where deflections, strains or accelerations are measured, can be used. Unfortunately, in practice inspection results are often not used directly to improve our knowledge of the degree of deterioration and it is difficult to combine information from different types of tests when making inferences. In this contribution, a methodology is developed and demonstrated in order to update estimations of parameters of service life models for concrete girders subjected to chloride-induced corrosion based on heterogeneous measurement data, with focus on strains measured under proof-loading and modal parameters extracted from ambient vibration tests. A Bayesian framework is adopted, where posterior distributions of the parameters describing the corrosion process are generated based on Markov Chain Monte Carlo sampling. These updated distributions reflect the information on the actual deterioration state of the bridge, which can be extracted from limited data. The updating procedure based on strain and modal data is illustrated by application on a simply supported beam and a case study.

Reinforced concrete structures are often subjected to corrosion, inducing a decreasing reliability over time. This corrosion often exhibits spatial variation, which can be modelled by random fields. To improve the accuracy of the estimated reliability, measurements can be performed. In this work, focus will be on static strain data. Taking into account these measurements leads to updated distributions of the corrosion parameters and an updated reliability. However, modelling errors will arise due to an inadequate choice of the spatial correlation model. These are present since the perfect model is not known. By assuming an inaccurate correlation model, a bias on the distribution of the reliability index might arise. In this work, the influence of this modelling error is investigated by assuming a different correlation model in the calibration of the model and the generation of the data. Different correlation models are considered and the most robust spatial correlation model is searched for. This problem is considered first for a simply supported beam subjected to corrosion, with the aim of identifying the most robust model. Next, it is investigated whether this model also performs well for a bridge structure.

In this work, a reinforced concrete slab bridge (instrumented and tested in 2018) is investigated. Based on field data, a finite element model of the bridge is calibrated. Model selection is performed both based on log evidence and posterior predictive capabilities. It is investigated if the models selected based on the log evidence also induce the most accurate posterior predictions. The influence of different assumptions on modelling the spatial distribution of the stiffness and different possible suggestions on how to include prediction errors and model bias are investigated. Comparing the conclusions based on log evidence and posterior predictions, only using the log evidence for model selection could be debated. Models performing best when considering the log evidence led to the least accurate posterior predictions, and models rejected based on the log evidence could still have good predictive capabilities. Considering the different model classes, introducing spatial variation of the stiffness leads to a posterior prediction closer to the measurements. Introducing a global model bias leads to a better match between predictions and measurements compared to not including this model bias. Even better posterior predictions are achieved if this model bias is quantified locally for the different considered datapoints.