Reconstruction of missing information in water distribution networks based on graph theory

Urban water infrastructures (UWIs), specifically water distribution networks (WDNs) are critical assets in cities affecting the well-being and safety of residents. There are challenges associated with data availability of these infrastructures, leading to the lack of detailed data required for different analysis aspects. The reason is that UWIs are often expected to have a lifetime of up to 100 years or more. Hence, this makes it difficult to access high-quality historical data needed for various modeling aims such as rehabilitation and modeling of network transition or deterioration. In addition, due to poor data management in water utilities, network information is often missing or only available in insufficient quality. Therefore, an automatic reconstruction approach is needed to rebuild missing data of UWIs with partly available data. To address this issue, a novel procedure is developed in this work based on complex network theory and using tailored graph measures (e.g., dynamic edge betweenness centrality) to reconstruct information. As a showcase, missing diameter information is systematically and successively reconstructed in WDNs based on graph characteristics of available diameters, street network data, and neighboring pipes. To validate our approach, a real-world WDN is used as a case study, where their pipe diameters are gradually deleted to increase the data gap. Afterward, the structural (i.e., diameter) and hydraulic (i.e., pressure) features of the reconstructed networks are compared with the existing network. The results show that a significant part of unknown diameters can be successfully reconstructed until a certain extent of the data gap.