ABSTRACT
The article demonstrates the climate change impact on the corrosion damage of an inspected bridge abutment (6 meters by 1.5 meters) through the maps of corrosion alarm probabilities obtained from a Bayesian network (BN) predictive model. The methodology for establishing corrosion alarm mapping of the first layer reinforcements consists of three steps: building the database, learning the BN structure and parameters from the non-destructive inspection data (NDT data), and using the BN for creating the corrosion alarm map. The studied BN model has four 4 variables: corrosion potential, corrosion rate, electrical resistivity of concrete cover, and corrosion alarm (related to corrosion rate). The available corrosion potential values (NDT data) of the inspected bridge abutment are used as the new information for updating the BN model to obtain the probability corrosion alarm map. Two scenarios are proposed in the study: decrease of 50 mV in all points, and decrease of 50 mV in the middle part, and of 100 mV in the edges. The results open a perspective to implement the real environmental exposition classes of the structures in the BN model to obtain a resilient maintenance (for corrosion of reinforcements) strategy.
