Bayesian inference of pit corrosion in prestressing strands using Markov Chain Monte Carlo method

Abstract

Corrosion monitoring has been widely studied to maintain the structural capacity of bridges through direct visual and nondestructive inspections or indirect inverse analysis-based methods. This study proposes a Bayesian inference method for inferring pit corrosion in the prestressing strands of prestressed concrete (PSC) bridges, which is an indirect method for corrosion monitoring. First, the probabilistic relationship between the mechanical properties of the strands and the amount of pit corrosion was defined using Bayes’ rule. Subsequently, a Markov chain Monte Carlo method was introduced to infer the posterior probability, which is a conditional probability distribution of the amount of corrosion given a certain mechanical property. Based on the inference results, probabilistic bounds for the amount of corrosion were derived. The proposed method was applied to two examples: (a) probabilistic corrosion inference of strands based on the tensile test results, and (b) probabilistic corrosion inference of embedded strands in PSC girders based on the bending test results. The inference results demonstrated the applicability of the proposed method.