Exploratory study on experimental and prediction based stress-strain behaviour of corroded prestressing steel strands

Prestressing steel strands have been extensively used in recent years due to the increased popularity of prestressed concrete structures. However, the corrosion of strands has a detrimental effect on these structures. This necessitates a detailed investigation of the mechanical properties of the corroded strands and to understand their influence on the strength of these structures. Therefore, in this paper, a systematic attempt has been made to study the mechanical properties of corroded steel strands corroded using the accelerated corrosion method. Initially, the tensile test was conducted on these corroded steel strands to study the mechanical properties of the strands and X-ray diffraction analysis was done on the corrosion products of these corroded steel strands. Then, resilient backpropagation with backtracking neural network based prediction models were developed to predict the ultimate strength of the corroded strands. The experimental results showed that an increase in the corrosion level reduces the ultimate stress, yield stress, breaking stress and the corresponding strains of the strand. The proposed prediction models give optimum results of cost functions for ultimate stress and ultimate strain predictions. The comparison study of ultimate stress-strain results obtained from the proposed prediction models showed remarkable performance with the experimental results.