Bayesian probabilistic model for reinforcement corrosion ratio of reinforcement in concrete prediction based on modified half-cell potential

In this study, a probabilistic model for predicting the corrosion rate of reinforcement in concrete by nondestructive half-cell potential testing was established. Half-cell potentials corresponding to different corrosion ratios, moisture contents, temperatures, water–binder ratios, and safeguard depth conditions were all measured in an accelerated corrosion test. Subsequently, stepwise regression and Bayesian estimation methods were conducted to establish a probabilistic model for predicting the corrosion rate of reinforcement in concrete. The experimental results showed that the corrosion ratios of reinforcements in concrete were inversely proportional to the measured half-cell potential. In contrast, the measured results of the half-cell potentials of reinforcements with identical corrosion ratios might decrease with increasing moisture content and temperature. The accuracy of the proposed probabilistic prediction model that considers the influence of moisture content and temperature was confirmed by comparing the prediction results with experimental data in this study and published literature. Finally, the effects of moisture content and temperature on the corrosion rate predictive accuracy were analyzed by comparing different probabilistic prediction models that only considered the correction of moisture content or temperature. The analysis results showed that the moisture content in concrete contributed more to the prediction accuracy of the reinforcement corrosion ratio than the testing temperature.