Journal of Water Resources Planning and Management

Forum papers are thought-provoking opinion pieces or essays founded in fact, sometimes containing speculation, on a civil engineering topic of general interest and relevance to the readership of the journal. The views expressed in this Forum article do not necessarily reflect the views of ASCE or the Editorial Board of the journal.

Forum papers are thought-provoking opinion pieces or essays founded in fact, sometimes containing speculation, on a civil engineering topic of general interest and relevance to the readership of the journal. The views expressed in this Forum article do not necessarily reflect the views of ASCE or the Editorial Board of the journal.

Water distribution system (WDS) model calibration research has focused on estimating model input/output parameters and analyzing several uncertainties (e.g., model uncertainty) to improve models with best-fit parameters. Numerous studies have shown that optimization algorithms generally quickly converge to very good parameter solutions. However, the generality and reasoning behind this have not been identified. This paper examines the shape and convexity of WDS response surfaces (i.e., objective function surfaces) and whether the surfaces have single global or multiple local optima. To that end, three networks with different network topologies are evaluated: (1) the Modena network as presented, (2) a modified form of the Modena network, and (3) a real Austrian network. Various conditions were evaluated to consider field measurement error, parameter uncertainty through pipe grouping, and model uncertainty. Results demonstrate that the response surfaces remained smooth and convex even when uncertainties are introduced, but the best parameter solutions are shifted from the true solution. The impact and sensitivities of the uncertainties are evaluated by examining the change in best-fit parameter estimates.

Water utilities are challenged to reduce their water losses through detecting, localizing, and repairing leaks as quickly as possible in their aging distribution systems. In this work, we solve this challenging problem by detecting multiple leaks simultaneously in a water distribution network for the Battle of the Leak Detection and Isolation Methods. The performance of leak detection and localization depends on how well the system roughness and demand are calibrated. In addition, existing leaks affect the diagnosis performance unless they are identified and explicitly represented in the model. To circumvent this chicken-and-egg dilemma, we decompose the problem into multiple levels of decision-making (a hierarchical approach) where we iteratively improve the water distribution network model and so are able to solve the multileak diagnosis problem. First, a combination of time series and cluster analysis is used on smart meter data to build patterns for demand models. Second, point and interval estimates of pipe roughnesses are retrieved using least squares to calibrate the hydraulic model, utilizing the demand models from the first step. Finally, the calibrated primal model is transformed into a dual model that intrinsically combines sensor data and network hydraulics. This dual model automatically converts small pressure deviations caused by leaks into sharp and localized signals in the form of virtual leak flows. Analytical derivations of sensitivities with respect to these virtual leak flows are calculated and used to estimate the leakage impulse responses at candidate nodes. Subsequently, we use the dual network to (1) detect the start time of the leaks, and (2) compute the Pearson correlation of pressure residuals, which allows further localization of leaks. This novel dual modeling approach resulted in the highest true-positive rates for leak isolation among all participating teams in the competition.

To detect leakage in urban water distribution networks and study the relationship between monitoring information and leakage diagnosis, the XGBoost algorithm was applied to identify the leakage zone and predict the leakage level. Software was adopted to call EPANET V2.2 for analyzing a water distribution network model, and emitters were added in the middle of pipes to simulate leakage events. By changing the discharge coefficient, the leakage flow was varied in hydraulic analysis. The node pressure sensitivity matrix was calculated, and the sensor placement and pipe zones were determined using the fuzzy $c$-means clustering method. Different leakage scenarios were simulated, and the location and level of leakage were identified through pressure changes at monitoring points based on the XGBoost algorithm. Taking two hydraulic models of water distribution network as examples to simulate and predict, which were compared with back-propagation neural network algorithm, it was revealed that the XGBoost algorithm can not only identify the leakage zone, but also predict the leakage level well. On the basis of sensor placement by Fuzzy c-means algorithm, for different leak scenarios, the average identification accuracy of the XGBoost algorithm was 5.54% higher than that of the back-propagation neural network algorithm in leakage zone. The average prediction accuracy of the XGBoost algorithm was 2.71% higher than that of the back-propagation neural network algorithm in leakage level. The XGBoost algorithm effectively can identify the leakage of water distribution networks.

Electric power utilities face a wide range of risks that cause financial uncertainty, with potential impacts on prices for customers. Among these, weather variability and retail load defection are perhaps two of the most studied. Whereas weather extremes expose utilities to unpredictable swings in electricity supply and demand, retail load defection, in which customers adopt distributed energy resources or switch to alternative providers, can alter utility business models fundamentally. We showed for the first time that these two phenomena can interact dynamically, with potential negative consequences for electricity ratepayers. We found that retail load defection could alter utilities’ financial exposure to weather risk in a matter of years. Using open-source power system simulation software coupled with a utility financial model, we simulated outcomes for a major hydropower-producing California electric utility under stationary hydrometeorological uncertainty, and under three different load defection scenarios ranging from 0% to 90%. We found that as load defection increases, the utility’s three main businesses (wholesale generation, transmission, and retail distribution) shift in relative importance. As a consequence, the impacts of interannual variability in hydropower production and demand in the utility’s system (a function of streamflow and air temperatures, respectively) become significantly altered. Air temperatures (a proxy for demand) become more predictive of utility financial performance, whereas the utility’s exposure to hydrology is poised to shift in complex ways. Drought will remain a major risk, but extremely wet years (which historically are beneficial to the utility’s significant hydropower holdings) may become damaging due to their association with low market prices. Our results also suggest that load defection could put much more pressure on utilities to make major annual rate adjustments or quickly adapt current strategies for managing weather risk.

This paper presents a novel algorithm with improved convergence and robustness for the pressure-driven modeling of water distribution networks (WDNs), to be implemented as hydraulic engine in the fourth release of the SWANP version 4.0 software. The innovative approach is based on increasing the order of convergence, which is quadratic for algorithms obtained from the Newton Raphson linearization of the equations for WDN resolution. As an example, the cubic order of convergence is obtained by evaluating system matrices at the generic iteration in a more refined way to account for the curvature of the hyperplane associated with the system in the direction of the Newton Raphson step. To show the benefits of the methodology, a third-order algorithm is constructed and compared with a traditional second-order. Both algorithms are based on the direct pressure-driven formulation expressing outflows as a function of service pressure and are equipped with the dampening of the Newton Raphson step. Applications on two case studies of different size, in which challenging pressure-driven conditions are created through demand amplification and segment isolation scenarios, prove that the methodology always reduces the total number of iterations required for convergence and the application of the step dampening. Overall, the results also show that the more stable convergence behavior is accompanied by an appreciable reduction in computation times. Further analyses proved that the third-order algorithm has similar convergence properties to algorithms based on the inverse pressure-driven formulation recently proposed in the scientific literature and can therefore be considered as a valid alternative to these algorithms.

Water losses through leakage represent a significant problem for asset management in water distribution systems. The interpretation of fluid transient pressure waves after the generation of a transient event has been previously used as a technique to locate and characterize leaks, but existing approaches are often both model-driven and limited to the existing knowledge of the system. The potential of using artificial neural networks (ANN) and fluid transient waves to detect, locate, and characterize anomalies in water pipelines has recently been proposed. However, its application in more realistic conditions (e.g., in the presence of background pressure fluctuations) has proven challenging. To address this, one alternative to enhance the response of any nonlinear system includes the introduction of artificial noise, a phenomenon known as stochastic resonance. In this paper, the enhanced detection of leaks in pressurized pipelines via the deployment of stochastic resonance is demonstrated. This paper harnesses this approach by presenting a methodology for the active inspection of pipelines using convolutional neural networks (CNNs). This methodology finds the optimal artificial noise intensity to be introduced into the training dataset for a set of CNNs. The methodology has been applied to a real pipeline in a laboratory at the University of Adelaide in which 14 transient experimental tests were conducted. The results indicated that the addition of noise to the transient pressure head training samples significantly enhances the CNN predictions for the leak location highlighting the existence of an optimum noise intensity to obtain both accurate and reliable results. When trained with the optimum noise intensity, the CNNs were able to locate leaks with an average error of 0.59% in terms of the actual location (in a 37.24-m long pipeline), demonstrating the promising potential of developing techniques based on CNNs to detect leaks and anomalies in water pipelines.

The economy continues to grow in Taiwan, and water demand from both households and industry has also continued to increase, which has led to water scarcity problems rising throughout the island. Thus, groundwater plays an essential role as a backup to supplement water resources. Considering that overpumping might lead to groundwater depletion, land subsidence, and even water quality issues, drawing up a more efficient strategy for groundwater management is necessary. This study proposes a dynamic adjusting pumping rate based on the exceedance probability of the historical groundwater table and the regional sensitivity of the pumping rate. The sensitivity of regional groundwater depletion was calculated by solving a Jacobian matrix composed of various pumping scenarios from numerical simulations. The coastal region of the Taichung City of Taiwan was selected as the study site to test our method. The result indicated that the groundwater table drawdown affected by pumping was reduced from 3.0 to 0.17 m in the 2015 dry season. During the period, groundwater of approximately 750,000 t was provided. Similarly, 350,000 t of groundwater resources and only a 0.09 m drawdown were accomplished in the 2016 wet season. Thus, this study proposed a groundwater dynamic management rule based on the current water level that effectively improves the efficiency of using groundwater resources.

Freshwater resources in coastal areas are under intense pressure from excessive groundwater extraction, which amplifies saltwater intrusion (SWI) into coastal freshwater aquifers, such as in the Mekong Delta. Studies that combine socioenvironmental data and households’ psychological factors next to salinity measurement data to design groundwater conservation strategies are rare. In this study, these aspects are combined to explore their influence on the public willingness to conserve groundwater using a Bayesian belief network model. We analyzed 313 household survey data spread over three districts in the coastal province of Tra Vinh, located in the Vietnamese Mekong Delta. The level of salinity is significantly correlated with the willingness to conserve groundwater. The top three socioenvironmental characteristics that influence willingness are the level of salinity, type of employment—i.e., being a farmer—and frequency of being exposed to groundwater or SWI promotional activities. Social norm, i.e., perceived social pressure, is the most influential psychological factor that determines willingness. This study reveals an urgency for the local government to intervene and create social pressure regarding the issue.