Spatial variability identification of carbonation depth in concrete using Bayesian networks

Abstract

Accurate prediction of carbonation depth is crucial for evaluating the durability and service life of reinforced concrete structures. Traditional methods for assessing carbonation depth often involve destructive testing, which is both costly and time-consuming, and yields results with limited accuracy, thus restricting their practical applicability. To address these shortcomings, this research introduces a novel two-step procedure that leverages inspection data on concrete porosity and saturation degree to estimate carbonation depth. By integrating Bayesian networks and considering the influence of spatial variability, the proposed methodology aims to enhance prediction accuracy compared to existing techniques. The study comprehensively investigates the impact of various factors, including the use of individual or combined inspection data, spatial dependence, and inspection distance, on prediction performance. The findings demonstrate the effectiveness of the proposed approach in capturing complex interactions between concrete properties, carbonation depth, and spatial variability. This research contributes to the advancement of non-destructive evaluation methods for concrete structures and provides valuable insights for optimizing inspection strategies.