Journal of Water Resources Planning and Management

Reliability assessment is an integral component of the decision-making process in the planning, design, and operations of water distribution networks (WDNs). Two different approaches are used to evaluate the reliability of WDNs: topological and hydraulic. Operational data and hydraulic layout in normal and abnormal conditions are not usually available to allow the computation of the hydraulic reliability. In this paper, four topological graph metrics (betweenness, topological information centrality, eigenvector centrality, and principal component centrality) were considered. Performance of the four metrics was compared with simulation-based hydraulic reliability. The comparison shows that no single topological graph metrics approach can capture characteristics of the complex networks. Using a Bayesian belief network (BBN)–based data fusion technique, the four topological graph metrics were combined into a single metric. The BBN model allowed embedding of the hydraulic process and capturing the uncertainty related to demand fluctuations and flow pattern changes in the network. The approach is applied to the Richmond case study and the results identify the majority of vulnerable areas defined using the hydraulic model and provide the ranking of the priority of interventions in WDNs. A Spearman rank correlation analysis was undertaken, and a heat map of the different results were generated for visual observation. The result from the data fusion technique has significantly improved accuracy of the topological graph metrics.

The intrusion of a foreign substance into the water distribution system represents a serious threat to public health. Large-scale water distribution systems serve thousands of consumers who may be put at risk to exposure and ingestion of potentially harmful substances. For an authority managing a water distribution system, it is important to (1) detect a potential contamination, and (2) locate the point of intrusion. However, points of known water quality data are expected to be sparsely distributed throughout the water distribution system, and may not provide sufficient data to quickly and accurately localize a contamination event. In this work, an inline mobile sensor was employed for the contamination event localization task in a Bayesian framework, such that the water quality data acquired by the mobile sensor were used to update the contamination intrusion location probabilities in the water distribution system. Using the Bayesian localization method was shown to improve the localization accuracy of a contamination event, with substantial improvements in the precision of localization.

This study presents a regional, probabilistic framework for estimating streamflow via spatial scaling for the purposes of informing water balance estimates in the Great Lakes basin. The framework follows a twofold strategy, including a quadratic programming–based optimization model to explore model structure, and several variants of a hierarchical Bayesian model based on insights found in the optimization model. The Bayesian model is developed to explore three innovations in hierarchical modeling for reconstructing historical streamflow at ungauged sites: (1) the combination of spatial proximity and physical basin characteristics for streamflow scaling, (2) the use of time-varying weights in the spatial scaling based on climate information, and (3) the consideration of residual heteroscedasticity to improve estimates of uncertainty. The proposed model variants are validated in a cross-validation framework to confirm specific hypotheses embedded in the model structure. Results suggest that each of the three innovations improve historical out-of-sample streamflow reconstructions and estimates of uncertainty around the reconstructed values, with the greatest improvements coming from the combined use of spatial proximity and physical similarity in estimating scaling weights at multiple donor sites.

Dividing a water distribution network into subsystems can improve the efficiency and ease of achieving management goals. Subsystems or district metered areas (DMAs) are isolated control zones with a defined number of entrances and exits. This research presents an automatic approach based on graph theory, optimization, and a heuristic methodology to design DMAs for water systems by minimizing the coefficient of variation of demand similarity among DMAs. The multiphase procedure to design DMAs (MPPDMA) satisfies constraints specified to limit the number of entrances per DMA, meet maximum and minimum pressure at nonzero demand nodes, and maintain water levels of the tanks over extended periods of simulation. MPPDMA is applied to four water networks. The results demonstrate that MPPDMA can improve demand similarity among DMAs.

Evaluating the suitability of data from global climate models (GCMs) for use as input in water supply models is an important step in the larger task of evaluating the effects of climate change on water resources management such as that of water supply operations. The purpose of this paper is to present the process by which GCMs were evaluated and incorporated into the New York City (NYC) water supply’s planning activities and to provide conclusions regarding the overall effectiveness of the ranking procedure used in the evaluation. A suite of GCMs participating in Phase 3 of the Coupled Model Intercomparison Project (CMIP3) were evaluated for use in climate change projections in the watersheds of the NYC water supply that provide 90% of the water consumed by NYC. GCM data were aggregated using the seven land-grid points surrounding NYC watersheds, and these data with a daily timestep were evaluated seasonally using probability-based skill scores for various combinations of five meteorological variables (precipitation, average, maximum and minimum temperatures, and wind speed). These are the key variables for the NYC water supply because they affect the timing and magnitude of water, energy, sediment, and nutrient fluxes into the reservoirs as well as in simulating watershed hydrology and reservoir hydrodynamics. We attempted to choose a subset of GCMs based on the average of several skill metrics that compared baseline (20C3M) GCM results to observations. Skill metrics for the study indicate that the skill in simulating the frequency distributions of measured data is highest for temperature and lowest for wind. However, our attempts to identify the best model or subgroup of models were not successful because we found that no single model performs best when considering all of the variables and seasons.