An analytical approach to predict fatigue strength of corroded steel plates using random field-based corrosion surface modeling

Abstract

This study introduces a probabilistic method to predict the fatigue strength of corroded steel members by incorporating uncertainties related to corrosion depth and surface morphology using random field modeling. The modeling successfully captured the probability density of the target corrosion depth, accurately representing corrosion geometric characteristics. Validation against fatigue testing on corroded rolled beams showcased the method’s accuracy across a range of stress levels. The application of the proposed method for time-variant fatigue strength predictions unveiled a rapid decline in both fatigue constant A and the constant amplitude fatigue threshold (CAFT) during the initial stages of corrosion. In marine and urban environments, fatigue strength fell rapidly within the first 3 years after corrosion initiation, whereas in rural environments, fatigue strength was anticipated to start to decline gradually after 5 years. These projections underscore the necessity of regular inspections to assess the condition of protective coatings. Moreover, the degradation curves generated by this method can serve as a foundational element in establishing an effective maintenance strategy for structures.