Reliability life analysis of reinforced concrete structures (RCS) in salt corrosion environments: An application of the three-parameter Weibull distribution model

Understanding the degradation mechanisms and patterns of concrete is crucial for predicting the service life of concrete structures. Researchers have focused on establishing quantitative correlations between deterioration and time. Traditionally, Fick's second law and chloride diffusion models have been key in durability design. However, the three-parameter Weibull distribution, widely used in reliability analysis, has been underutilized in assessing reinforced concrete durability. To address this, this study investigates the corrosion degradation behavior of reinforced concrete exposed to a coupled salt solution environment (0.32 mol/L NaCl and 0.4 mol/L MgSO₄), utilizing electrochemical testing and a three-parameter Weibull reliability model. The electrochemical results indicated a gradual increase in the corrosion current density, reaching a critical threshold of 10 μA·cm⁻² after 720 days, marking the onset of significant corrosion. The predicted reliability lifespan of C35 reinforced concrete under these aggressive conditions was estimated to be approximately 760 days. To estimate the Weibull parameters (shape parameter U, scale parameter V, and location parameter δ), three parameter estimation methods were employed: the Correlation Coefficient Optimization Method (CCOM), Maximum Likelihood Method (MLM), and Method of Moments (MEM). Among these, CCOM and MLM provided consistent and stable estimates, while MEM exhibited larger estimation errors. The reliability curves derived from CCOM and MLM displayed similar trends, with reliability maintaining a constant value in the initial phase and then rapidly decreasing after 730 days. These findings not only deepen the understanding of reinforced concrete’s corrosion behavior in aggressive environments but also suggest that CCOM and MLM are the most reliable methods for parameter estimation in cases with limited sample sizes.