Abstract
Cracks in reinforced concrete structures can accelerate the local depassivation of reinforcement due to carbonation. Different approaches have been proposed to account for pre-existing cracks within engineering models to predict the carbonation depth. In this study, we provide a detailed comparative analysis of different extensions available for the fib carbonation model to account for cracks, viz., crack influence factor (CIF) approaches, a diffusion-based model and the crack depth adaption. The model extensions are first validated against a dataset of lab data collected from the literature and additional experiments performed as the part of this study. The CIF approaches achieved the highest accuracy for the carbonation depth prediction when compared against lab data. The diffusion-based model was inaccurate for low CO2 concentrations. The crack depth adaption provides overly conservative results. No model was found to be best performing, and large scatter was observed between predicted and experimental values. This emphasizes the need for more detailed multi-physics-based models to achieve accurate predictions. For further comparison, service life predictions were conducted for two structural scales, viz., the whole structure and the cracked area. It is concluded that the choice of model extension and the structural scale of analysis have a large influence on predicted probability of failure.