Abstract: The increasing amounts of pressure and threat on pipeline infrastructure consequently represent an elevation in the number of pipeline failures experienced. These failures are accompanied with extensive damage leading to environmental, social and economic stress to municipalities and water utilities. Respective managers are therefore pressured to put in place reliable maintenance and rehabilitation strategies in effort of minimizing losses. Prediction of potential mishap is one way through which instigation of planned rehabilitation may be upheld. However, this is challenging, thanks to inherent uncertainties. One effective way of handling uncertainty is through collection and combination of auxiliary information and knowledge which can be tackled using probabilistic models like Bayesian Networks (BNs). In this study, therefore we present comprehensive review of how probabilistic models have been applied in different ways to predict pipeline leakage; we identify various gaps presented by these models and finally we highlight the current state of research as far as leakage prediction is concerned. We also propose a recommendation for future research work.
G.A. Ogutu, P.K. Okuthe and M. Lall, 2017. A Review of Probabilistic Modeling of Pipeline Leakage using Bayesian Networks. Journal of Engineering and Applied Sciences, 12: 3163-3173.