Assessment of Infrastructure Reliability in Expansive Clays Using Bayesian Belief Network

Abstract

Civil infrastructure supported by expansive clays is severely affected by extensive volumetric deformations. The reliability prediction of such facilities is quite challenging because of the complex interactions between several contributing factors, such as a scarcity of data, a lack of analytical equations, correlations between quantitative and qualitative information, and data integration. The main contribution of this research is the development of a modeling approach based on the Bayesian belief network. The modeling results highlight that facility age is the most critical parameter (23% variance), followed by facility type (1.37% variance), for all the investigated types of infrastructure, namely road embankments, buried pipelines, and residential housing. Likewise, the results of sensitivity analysis and extreme scenario analysis indicate that the new method is capable of predicting infrastructure reliability and the assessments were found to be in agreement with expected field behavior. The proposed model is useful in decision making related to civil infrastructure management in expansive clays.