Evaluating and enhancing resilience in water infrastructure is a crucial step towards more sustainable urban water management. As a prerequisite to enhancing resilience, a detailed understanding is required of the inherent resilience of the underlying system. Differing from traditional risk analysis, here we propose a global resilience analysis (GRA) approach that shifts the objective from analysing multiple and unknown threats to analysing the more identifiable and measurable system responses to extreme conditions, i.e. potential failure modes. GRA aims to evaluate a system's resilience to a possible failure mode regardless of the causal threat(s) (known or unknown, external or internal). The method is applied to test the resilience of four water distribution systems (WDSs) with various features to three typical failure modes (pipe failure, excess demand, and substance intrusion). The study reveals GRA provides an overview of a water system's resilience to various failure modes. For each failure mode, it identifies the range of corresponding failure impacts and reveals extreme scenarios (e.g. the complete loss of water supply with only 5% pipe failure, or still meeting 80% of demand despite over 70% of pipes failing). GRA also reveals that increased resilience to one failure mode may decrease resilience to another and increasing system capacity may delay the system's recovery in some situations. It is also shown that selecting an appropriate level of detail for hydraulic models is of great importance in resilience analysis. The method can be used as a comprehensive diagnostic framework to evaluate a range of interventions for improving system resilience in future studies.

Accounting for water in a distribution system can be improved by dividing systems into smaller, metered zones. This paper proposes an approach that could create boundaries for district metered areas (DMA) automatically on the basis of the community structure of water distribution systems. Community structure—the gathering of vertices into communities such that there is a higher density of edges within communities than between them—is a common property of many complex systems. For verification, the method was tested on a real-world distribution system, and the result was compared with a manually designed DMA layout. Although further improvements are necessary, because the achieved community structure is in excellent agreement with the zoning plan in reality, this approach is a new addition to the number of automated methods aimed at complementing and eventually substituting the empirical trial-and-error approach.

Improving the resilience of water distribution systems (WDSs) to handle natural disasters (e.g., earthquakes) is a critical step toward sustainable urban water management. This requires the water utility to be able to respond quickly to such disaster events, and in an organized manner, to prioritize the use of available resources to restore service rapidly while minimizing the negative impacts. Many methods have been developed to evaluate the WDS resilience, but few efforts are made so far to improve the resilience of a postdisaster WDS through identifying optimal sequencing of recovery actions. To address this gap, the authors propose a new dynamic optimization framework in this study in which the resilience of a postdisaster WDS is evaluated using six different metrics. A tailored genetic algorithm is developed to solve the complex optimization problem driven by these metrics. The proposed framework is demonstrated using a real-world WDS with 6,064 pipes. Results obtained show that the proposed framework successfully identifies near-optimal sequencing of recovery actions for this complex WDS. The gained insights, conditional on the specific attributes of the case study, include the following: (1) the near-optimal sequencing of a recovery strategy heavily depends on the damage properties of the WDS; (2) replacements of damaged elements tend to be scheduled at the intermediate-late stages of the recovery process due to their long operation time; and (3) interventions to damaged pipe elements near critical facilities (e.g., hospitals) should not be necessarily the first priority to recover due to complex hydraulic interactions within the WDS.

Water quality sensors placed in water distribution systems (WDSs) are critical for detecting accidental or intentional contamination intrusion. This motivates research to optimally place a limited number of sensors for a given WDS aimed to maximize the detection effectiveness (e.g., the detection likelihood and time to detection). Typically, effectiveness of a sensor placement strategy (SPS) is assessed using the expected impact across a number of contamination scenarios. Despite the value of such an approach, it may provide limited information on the SPS’s comprehensive properties in detection, such as the SPS’s ability to detect events with different levels of consequences, or to reduce the impacts from undetectable events. To address this limitation, this study investigates the underlying characteristics of the SPS’s effectiveness for contamination detection using a set of metrics focusing on detection time, consumption of contaminated water, the number of contaminated demand nodes, and the contaminated spatial distance. The former two are derived from detectable contamination scenarios and the latter two are computed from both detectable and undetectable contamination scenarios. The proposed method is illustrated for two real-world WDSs, and the results reveal the underlying properties of the SPS’s utility in contamination detection, which is significantly more informative than the measure of expected impacts. Such improved understanding provides guidance for selecting the most appropriate SPS and for improving preparedness for contamination events.

This paper introduces a multi-objective optimisation approach for the challenging problem of efficient sensor placement in water distribution networks for contamination detection. An important question is how to identify the minimal number of required sensors without losing the capacity to monitor the system as a whole. In this study, we adapted the NSGA-II multi-objective optimisation method by applying centrality mutation. The approach, with two objectives, namely the minimisation of Expected Time of Detection and maximisation of Detection Network Coverage (which computes the number of detected water contamination events), is tested on a moderate-sized benchmark problem (129 nodes). The resulting Pareto front shows that detection network coverage can improve dramatically by deploying only a few sensors (e.g. increase from one sensor to three sensors). However, after reaching a certain number of sensors (e.g. 20 sensors), the effectiveness of further increasing the number of sensors is not apparent. Further, the results confirm that 40–45 sensors (i.e. 31 − 35% of the total number of nodes) will be sufficient for fully monitoring the benchmark network, i.e. for detection of any contaminant intrusion event no matter where it appears in the network.