Failure Event Probability Calculation in Wastewater Collection Systems Using the Bayesian Network

Wastewater systems form an important urban infrastructure that are used for the collection and treatment of wastewater for return into the environment or water reuse. The sewers network in this system forms its most important component, any failure in which may lead to adverse consequences and disruption in urban life. Proper functioning of a sewers network depends on its operation and maintenance (O&M) program that requires timely inspections to identify the high risk sewers with any likelihood of failure in order to gurantee the sustained and sound performance of the whole network. In this study, the Bayesian network is used to develop a model for calculating failure event probability in wastewater collection systems. Given the capabilities of the Bayesian network and the characteristic features of the sewers network, the proposed model is capable of predicting likely failure events. The procedure used for model implementation consists of the following four main steps: preparation of model inputs, training the Bayesian network, validation of the trained network, and receiving output results. To illustrate the application of this method, part of Tehran wastewater collection system is selected as a case study and failure probabilities are calculated. Based on the model results, Tehran sewers can be divided into five categories priotized according to inspection and maintenance requirements. The results indicate that the probability of failure for most of the existing sewers is very low and low (37%) or moderate (60%) due to the newly annexed sewers in the collection system.