Nowadays, optimal sensor placement (OSP) for leakage detection in water distribution networks is a lively field of research, and a challenge for water utilities in terms of network control, management, and maintenance. How many sensors to install and where to install them are crucial decisions to make for those utilities to reach a trade-off between efficiency and economy. In this paper, we address the where-to-install-them part of the OSP through the following elements: nodes’ sensitivity to leakage, uncertainty of information, and redundancy through conditional entropy maximisation. We evaluate relationships among candidate sensors in a network to get a picture of the mutual influence among the nodes. This analysis is performed within a multi-criteria decision-making approach: specifically, a herein proposed variant of DEMATEL, which uses fuzzy logic and builds comparison matrices derived from information obtained through leakage simulations of the network. We apply the proposal first to a toy example to show how the approach works, and then to a real-world case study.

This work explores the various relationships between several parameters in an incident hub of a small real WDN. To facilitate analysis, incident data were previously categorised depending on the nature of the cause of the incident. The data are analysed through spatial analysis, based on the transmission of information from the surrounding areas, and are also incorporated to add certain uncertainty to the information. The results of this characterisation are presented and analysed in this contribution. The results are promising in providing water distribution networks with key parameters that promote prediction and classification processes from the lessons learned.