Probabilistic prediction of ultimate strength and strain of corroded steel strands

Abstract

Steel strands are widely used in civil infrastructure owing to their high tensile strength and bending flexibility. However, they are vulnerable to corrosion; thus, it is important to predict the ultimate strength and strain of corroded strands. Although various studies have been conducted for this purpose, it remains difficult to make precise predictions owing to the unique structural complexity of steel strands and various sources of uncertainty. This study proposes a new method for the probabilistic prediction of the ultimate strength and strain of corroded steel strands. First, finite element (FE) models are constructed for several types of corroded wires. Second, according to the FE analysis results, a surrogate model is developed using Gaussian Process regression. Third, employing the Monte Carlo simulation with a theoretical strand model, the ultimate strength and strain of a corroded steel strand are predicted probabilistically. In the proposed method, the Monte Carlo approach enables providing 95% and 99% bounds for the ultimate strength and strain of a corroded steel strand. The proposed method is applied to several specimens of corroded seven-wire strands, and the prediction results show good agreement with the test results.