Abstract: Among the primary durability threats for reinforced concrete structures is carbonation-induced corrosion of reinforcement. Despite significant research efforts in the last decades, models for carbonation ingress are associated with large case-specific model uncertainties. Therefore, this study compares selected predictive models for carbonation of reinforced concrete. Three representative frameworks are examined: the empirical power-law formulation of the fib Bulletin 112, the resistance-based approach of the fib Bulletin 34, and the semi-mechanistic model. Their performance is assessed in a case study on an existing cooling tower, which provides field data for calibration and validation. Deterministic analyses demonstrate that the different models yield consistent predictions, while probabilistic evaluation high-lights the dominant influence of concrete cover and the significant roles of diffusion parameters and model uncertainty. The comparative results confirm that reliability-based approaches provide a more robust basis for durability design than purely deterministic evaluations, with case-specific updating of key parameters proving essential for accurate service-life assessment. On average, both the resistance and semi-mechanistic models seem to reproduce field carbonation depths with reasonable accuracy (11–12 mm versus a measured mean of 14.8 mm). Probabilistic reliability analysis then indicates that the durability target of βt = 1.5 is reached after about 15 years of exposure.
Šomodíková et al. (Fri,) studied this question.