Influence of Bayesian Network Structure on Corrosion Risk Assessment of Steel Reinforcement in Concrete

Four Bayesian network models (SB-1, SB-2a, SB-2b, SB-2c) for assessing reinforcement corrosion probability in concrete were compared. Seven corrosion parameters were considered: potential; resistivity; chlorides; pH; corrosion current; charge transfer resistance; cracking. Monte Carlo simulations for three threat scenarios (low, medium, high) demonstrated the superiority of model SB-2c – highest accuracy and flexibility. Model SB-2c was identified as optimal for corrosion diagnostics of reinforced concrete.