Probabilistic capacity models for corroding posttensioning strands calibrate using laboratory results

The presence of air voids, moisture, and chlorides inside tendons or ducts was cited as a reason for the early age strand corrosion and failure in the Mid-bay, Sunshine Skyway, and Niles Channel posttensioned (PT) bridges in Florida, United States. Although rare, these incidents call for frequent inspection and structural reliability assessment of PT bridges exposed to moisture and chlorides. This paper develops and presents probabilistic strand capacity models that are needed to assess the structural reliability of such PT bridges and recommends a time frequency of inspection. A total of 384 strand test specimens were exposed to various void, moisture, and chloride concentration conditions for 12 and 21 months; the remaining tension capacities were then determined. Using this experimental data and a Bayesian approach, six probabilistic capacity models were developed based on the void type. The mean absolute percentage errors of these models are less than 4%, indicating that reasonably accurate prediction of the strand capacity is possible, when void, aggressive moisture, and chloride conditions are present.