In the event of a disruption of operation, parts of the water distribution networks (WDNs) must be temporarily disconnected from the supply source by the closure of isolation valves to allow pipe repair. For cost reasons, however, the number of such valves is usually limited, requiring strategies for their optimal placement. In this paper we combine graph theoretical approaches with reliability analysis by using the WDN topology and isolation valve information. A novel methodology for the assessment of valve placement strategies is developed, in which we investigate WDNs in their information space by means of complex network analysis. Unlike traditional approaches, we use the dual representation of the network, where WDN segments (i.e., a set of pipes) are considered as nodes and isolation valves as edges. With the developed algorithm, the WDNs are analyzed on the basis of the dual graph, providing new insights beyond a conventional graph (primal mapping) analysis. The method is applied to two real-world systems, to identify different patterns with respect to the probability density functions of (dual) node properties: node-degree P(k), aggregated pipe length P(l), and demand P(d). Additional complex network metrics, such as the characteristic path length, degree correlation, and modularity are investigated and discussed. The observed topological differences also reflect the availability of financial resources and the different types of water supply of the systems. The implications of the results allow for a novel assessment of WDN reliability and robustness in the event of disruption and network isolation.