Probabilistic inference on corrosion configurations based on tensile test results of corroded strands

Abstract

Prestressing strands are frequently affected by pitting corrosion. A new method was recently proposed to evaluate the effects of corrosion on the mechanical characteristics of strands in a probabilistic manner. By employing the proposed method inversely, we suggest a new inference method to examine corrosion variables in corroded strands, such as the pit depth and loss of sectional area, based on the tensile tests (i.e., ultimate load-displacement tests). The inference method requires solving an optimization problem using a cost function, which was defined as the difference between the experimental results and the predicted values from a previous study. As a result, corrosion configurations are inferred based on the pit configuration and tensile tests. The proposed inference method was verified when applied to three specimens of corroded strands from an existing prestressed concrete bridge in South Korea, and the relation between corrosion configurations and mechanical properties of corroded strands is discussed.