Evaluating Parameters Influencing the Emergence and Increase of Non-Revenue Water Using Bayesian Networks (Case Study: District 4 of Tehran Water and Wastewater Company)

Effective strategies for optimal management of water distribution networks (WDNs) along with financial resources are clearly identifying design and performance parameters affecting the emergence of non-revenue water and reducing the impact of parameters. In this study, using references and experiences of experts in the field of WDN, the factors that are involved in the development of non-revenue water components were identified. Then, to collect the required information on the status of the parameters, officials’ and experts’ opinion were asked through questionnaires. Bayesian Networks (BNs) were used as a modeling tool so as to not only consider uncertainties associated to the lack of sufficient data and detailed information on non-revenue water components but also consider the probabilistic relationships between parameters. Finally, in order to analyze BN models results, a sensitivity index was proposed to prioritize parameters based on their impact. To investigate the usefulness of the proposed model, the area covered by District 4 of Tehran Water and Wastewater Company was selected as a case study. The results indicated that the probability that apparent losses, real losses and non-revenue water are high is 37.48 %, 35.04% and 32.2%, respectively.