Prediction of Chloride Induced Corrosion Using Dynamic Bayesian Updating and Monitoring Data

Abstract

Corrosion of steel rebars represents the main cause of deterioration of reinforced concrete (RC). Corrosion of steel rebars occurs with the break of the passivation protective layer on the surface of rebars due to carbonation or chloride contamination of concrete cover. Estimating the risk of corrosion initiation in time can provide reliable information to protect rebars and plan maintenance of RC structures. This paper presents a framework for a probabilistic modelling of corrosion initiation in RC structures induced by chloride contamination. The approach is based on a dynamic Bayesian network updated over time using corrosion monitoring data. The dependencies between variables are represented in a directed graph and joint probability density functions are computed using Fick’s law and Monte Carlo sampling. The approach is illustrated with the CorroVolta® sensor, an embedded corrosion sensor designed and manufactured at Socotec Monitoring. Results prove the feasibility of using the dynamic Bayesian approach for corrosion monitoring of RC structures and would support a reliable decision basis for reinforced-concrete bridges and wind-turbine foundations.