Reliability Analysis of Corroded RC Structures Integrating Hybrid Bayesian Network

Abstract

The performance of RC structures might deteriorate with time due to environmental stressors (e.g., chloride ingress and concrete carbonation), compromising structural safety and social sustainability. Many studies have shown that probabilistic methods are necessary for the performance and reliability assessment of deteriorating RC structures, considering the uncertainty of environmental changes and material properties. However, most previous studies only considered the reliability prediction of deteriorated RC structures at the design stage. Recent studies proved that the inspection results could significantly reduce the uncertainties in the life cycle assessment of RC structures. Thus, a hybrid Bayesian network (HBN)-based reliability calculation framework is developed by integrating durability assessment and mechanical assessment of deteriorated RC structures. Experimentally validated 2D chloride transport models and analytical models are used to calculate the durability and mechanical properties of RC beams. Also, different dynamic Bayesian networks are combined into one HBN to perform Bayesian inference. In addition, a low-discrepancy pseudo-random sequence is utilized in building the HBN to improve the efficiency and accuracy of the modeling. Finally, to demonstrate the performance and detection results of the framework in the life-cycle reliability analysis of RC structures, the proposed framework is applied to an illustrative case.