Journal of Water Resources Planning and Management

Early detection of a contamination leach into a water distribution system, followed by the identification of the source and an evaluation of the total amount of the contaminant that has been injected into the system, is of paramount importance in order to protect a water user’s health. The ensemble Kalman filter, which has been recently applied in hydrogeology to detect contaminant sources in aquifers, is extended to the identification of a contaminant source and its intensity in a water distribution system. The driving concept is the assimilation of contaminant observations at the nodes of the pipeline network at specified time intervals until enough information has been collected to allow the positioning of the source and the estimation of its intensity. Several scenarios are analyzed considering sources at different nodes, with different delays between the beginning of the pollution and the start of the measurements, different sampling time intervals, and different observation ending times. The scenarios are carried out in the benchmarking Anytown network, demonstrating the ability of the ensemble Kalman filter for contaminant-source detection in real water distribution systems. The use of the ensemble Kalman filter supposed a major breakthrough in the inverse modeling of subsurface flow and transport, and the successful results of its application to the synthetic Anytown network warrant further exploration of its capabilities in the realm of water distribution systems.

Reservoir operations should consider both adaptiveness and robustness to deal with two of the main characteristics of climate change: nonstationarity and deep uncertainty. In particular, robust operational strategies are distinguished from risk-neutral expected value optimization in the sense that they should be satisfactory over a wider range of uncertainty and improve the ability of a reservoir system to adapt to climate change. In this study, a new framework named robust stochastic dynamic programming (RSDP) is proposed that couples robust optimization (RO) with the formulations of objective function or constraints used in stochastic dynamic programming (SDP). Two main approaches of RO, namely feasibility robustness and solution robustness, are both considered in the optimization algorithm. Consequently, this study uses the Boryeong multipurpose dam to evaluate three SDP framings: conventional-SDP (CSDP), RSDP-feasibility robustness (RSDP-F), and RSDP-solution robustness (RSDP-S). These three SDP formulations were used to derive optimal monthly release rules for the Boryeong Dam, and their relative performances were evaluated using simulations of a broader range of inflow scenarios. The simulation-based re-evaluations of the resulting reservoir operational policies were quantified using a wide range of metrics that include reliability, resiliency, and vulnerability, as well as regret-based robustness metrics. The results of this study suggest that the RSDP-S model not only increases the range of possible solutions, but also yields more desirable operation outcomes under extreme climate conditions with respect to both traditional and robustness metrics.

This work forwards time-varying sensitivity analysis of control policies as an approach to identify the value of the alternative timing of forecasts in a complex multipurpose reservoir system. The time-varying forecast diagnostics are demonstrated using the Conowingo Dam in the Lower Susquehanna River Basin. First, we evaluate the utility of multiple lead time forecasts, both in isolation and in combination, to determine the sensitivity of optimal operations to forecast lead time. We then use time-varying sensitivities to determine how the system utilizes a particular forecast through time for two policies that emphasize hydropower and environmental objectives, respectively, in order to determine how the value of inflow information changes with respect to stakeholder preferences and evolving hydrologic conditions. Results show that the value of inflow information varies greatly depending on the objectives being emphasized and can decline or fade entirely when system constraints (e.g., lower bounds on reservoir stage and spillway thresholds) inhibit the use of forecasts to inform decisions, particularly during extreme events.

The purpose of the paper is to present an approach for pipe burst detection, localization, and cross-sectional area quantification based on the changes in the discrete harmonic spectrogram and analysis of damping of fluid transients. The amplitude of each resonance response of the burst-induced transient wave in each analyzed time window in the spectrogram is damped differently because of the presence of the burst. A pressure signal processing algorithm for pipe burst detection and estimation has been developed by the authors to explore this specific type of analysis, in conjunction with a predefined gap between each window. Because the window gap can be set to be equal to the data transmission and sampling rates in the analysis to detect and estimate the burst, it can be applied during real-time data monitoring. Unlike other hydraulic transient-based detection methods, the proposed algorithm only requires the Nyquist frequency of the third resonance response as the data transmission and sampling rates, which are significantly lower than the same quantities in commonly used data acquisition systems. The algorithm has been verified both numerically and experimentally.

Appropriate sediment management is required to improve the lifespan and productivity of reservoirs. Flushing operations to release part of the trapped sediment have therefore become increasingly common in practice. However, there is no clear methodology or guideline to assess their effectiveness. Numerical models are promising, but the lack of studies and data for validation presents a significant challenge. An opportunity to assess flushing operations is presented by the increasing number of dam removal interventions monitored for subsequent reservoir erosion and sediment displacement. Treating dam removal as an extreme case of sediment flushing, we use the case of the Marmot Dam removal to validate a two-dimensional morphodynamic model, which is then used for the study and optimization of sediment flushing operations. The results indicate that the software can be successfully used to design and optimize these types of operations notwithstanding some limitations in the quantification of released sediment volumes. Flushing appears to be more efficient if it starts just before the peak of a flood wave.

Water management operations can drive long-term water quality patterns in nutrient-enriched lakes and reservoirs. This study identified multidecadal trends of rainfall, flow, water levels, total phosphorous (TP), total nitrogen (TN), and Chlorophyll-a in the water column and eight (dis)tributaries during four specific operation schedules (1974–1990, 1991–1999, 2000–2007, and 2008–2019) in Lake Okeechobee (Florida). We found an increasing trend in water discharges towards the Gulf of Mexico (via the Caloosahatchee River) during the dry season for the entire 46 years, and a decreasing trend in discharges towards the Atlantic Ocean (via the St. Lucie Canal) during 2008–2019. Higher flows, combined with increasing TP concentrations in the lake and the Caloosahatchee, led to significant increases in monthly TP mass loads to the Gulf of Mexico by up to 122%–158% during 2008–2019. Higher flows in the Caloosahatchee also resulted in a net TN load increase to the Gulf, despite system-wide decreases in TN concentrations. Overall, results from this study demonstrate that reservoir operations could have long-term effects on nutrient status and exports; thus, modifying operations should be considered as a potential nutrient management tool.

Most sampling design (SD) methods in water distribution networks are focused on the calibration of a hydraulic model and use local sensitivity analysis and demand-driven simulation methods to select sampling points. These SD methods have little adaptability for water quality models separately or with hydraulic models. In this paper, which presents a new approach, the Sobol method is used for sensitivity analysis. Then optimal points are selected based on sensitivity and cost criteria. It is possible to apply this method in a joint SD of hydraulic and quality models. This method is applied and verified in a case study. The effect of the head-driven simulation method (HDSM) and different parameters on SD are also evaluated. The results indicate that the Sobol method and HDSM cause a significant shift in the priority of some selected points. In addition, the proposed method for the joint SD of hydraulic and quality models has a suitable capability.

Automated detection and classification of sewer defects can complement the conventional labor-intensive sewer inspection process by providing an essential tool to classify sewer defects in a more efficient, accurate, and consistent way. This paper presents a convolutional neural networks (CNNs)–based model to automatically detect and classify six most commonly observed sewer defects (i.e., cracks, disjoints, obstacles, residential walls, tree roots, and normal categories) obtained from multisource CCTV images under various circumstances. Data augmentation techniques (including geometric and color transformations) are applied to enhance the model performance. The proposed CNN model is further compared with a state-of-the-art solution (retraining the SqueezeNet using defect images) by adopting transfer learning technique. An average prediction accuracy of 90% is achieved, indicating that the investigated defects can be well recognized by the model without any expert knowledge of sewer detection. There is a higher degree of confidence in predicting tree roots and disjoints, followed by residential walls and cracks. Results show that the prediction accuracy has increased by 15% thanks to data augmentation. Despite the transferred SqueezeNet model achieved a higher accuracy (95%), it cost almost 13 times the computation time of the CNN model. The study demonstrates the feasibility of the deep learning technology in the automated classification of sewer defects and advances the knowledge in the research field.

This paper presents an approach to managing surface water abstraction utilizing real-time flow forecasting and control techniques. To evaluate the effectiveness of alternative data-driven and process-based methods, flow forecasts at a case study site (River Dove, UK) using (1) a probability-distributed rainfall-runoff model (PDM), (2) PDM coupled with an autoregressive integrated moving average (ARIMA) error predictor, and (3) a long short-term memory (LSTM) neural network are integrated into a water resources management model coupled with genetic algorithm optimization to simulate and compare water abstractions, reservoir storage, downstream river flows, and pumping energy costs. When compared to historical data, results show that both PDM plus ARIMA and LSTM forecasts led to improved water abstraction operations, i.e., increased water abstraction volumes during dry periods while maintaining river environmental flows, as well as reduced pumping costs. Cost savings were found to be sensitive to the accuracy of the forecasting technique only within specific flow ranges. This study demonstrates the water resource benefits of real-time flow forecasting in supporting flexible water pumping schedules and further discusses the benefits of alternative modeling approaches in the specific context of controlling water abstraction.

This paper compares the economic performance of a dam construction strategy versus a no dam construction (i.e., the statu quo) strategy under multiple scenarios and using a multimodel ensemble of three microeconomic mathematical programming models. The result is a database of simulations representing multiple plausible futures which offers information on uncertainty regarding scenario assumptions and model structure (through the ensemble spread). Using an iterative robust decision-making framework, simulation results are coupled with experts’ knowledge and opinion to detect vulnerabilities in the proposed strategies, quantify potential trade-offs between responses, and identify a robust adaptation strategy. Methods are illustrated as applied to the Órbigo Catchment in Northwestern Spain, where the Douro River Basin Authority (DRBA) will decide in the coming months whether to build two dams to enhance irrigation water supply and reliability. Simulation results show that, for most scenarios and models considered, dam construction costs lead to water prices beyond the willingness of local cereal-growing irrigators to pay. Following a robust decision-making process, parties to the decision unanimously declared the status quo (i.e., no dam construction) strategy to be preferred to the dam construction strategy. This outcome substantiates the need for economic assessments of dam construction projects that account for nonlinear responses by agents to complement technical assessments in decision making.

Occurrence of high pressure in water distribution networks (WDNs) may generally result in leakage, bursting, and overconsumption. Temporal and spatial variations in pressure values cause dissatisfaction of consumers and distrust of WDNs. Pressure management is an operational strategy to improve the reliability of WDNs. This strategy should be considered as a cost-savings practice due to the fact that operating costs are a significant proportion of WDN costs. In the present study, the surplus pressure in a WDN, consumed by turbines, is used to generate energy and compensate for some of the costs. Due to the high cost and specialized construction of the turbines, the pumps working in reverse or pump-as-turbines (PATs) have been employed in the WDNs instead of turbines. This investigation aims to improve the network performance by installing PATs with the appropriate number, type, and time schedule in the right locations. In this regard, the optimal schedules for pumping and using pressure-reducing valves (PRVs) have been considered. A numerical model was developed to optimize consumers’ satisfaction and costs as objective functions. The efficiency of the proposed model was evaluated by selecting a case study with overpressure, and additionally, the objective functions were optimized by using two scenarios. The results associated with the first scenario indicated a reduction of network costs by up to 90% over its life cycle as well as an increase in network reliability. Through the second scenario, multiobjective optimization provided a trade-off diagram as the final solution for choosing various solutions with different cost levels and reliabilities.

Real-time water quality (WQ) sensors in water distribution networks (WDN) have the potential to enable network-wide observability of water quality indicators, contamination event detection, and closed-loop feedback control of WQ dynamics. To that end, prior research has investigated a wide range of methods that guide the geographic placement of WQ sensors. These methods assign a metric for fixed sensor placement (SP) followed by metric-optimization to obtain optimal SP. These metrics include minimizing intrusion detection time, thereby minimizing the expected population and amount of contaminated water affected by an intrusion event. In contrast to the literature, the objective of this paper is to provide a computational method that considers the overlooked metric of state estimation and network-wide observability of the WQ dynamics. This metric finds the optimal WQ sensor placement that minimizes the state estimation error via the Kalman filter for noisy WQ dynamics—a metric that quantifies WDN observability. To that end, the state-space dynamics of WQ states for an entire WDN are given and the observability-driven sensor placement algorithm is presented. The algorithm takes into account the time-varying nature of WQ dynamics due to changes in the hydraulic profile—a collection of hydraulic states including heads (pressures) at nodes and flow rates in links that are caused by a demand profile over a certain period of time. Thorough case studies are given, highlighting key findings, observations, and recommendations for WDN operators. Github codes are included for reproducibility.

The present study was designed to report experimental results for effective diffusion and longitudinal dispersion coefficients for both chlorine and fluoride added to water supply pipes. This is the first extensive study to find effective diffusion and dispersion coefficients for chlorine and fluoride added to water supply pipes. A pipe of field diameter and length was used to find dispersion coefficients for both chlorine and fluoride. The study covers the Reynolds number (R) from laminar and transitional flows ($\mathrm{R}=200–\mathrm{20,000}$), for which more correct information about the dispersion coefficients might be needed in the literature. The paper first presents the solution of a two-dimensional (2D) advection-diffusion-reaction equation of concentration movement in a pipe. Second, it presents the analytical solution of a one-dimensional (1D) advection-dispersion-reaction equation. Third, by using the numerical and analytical solutions, effective diffusion and longitudinal dispersion coefficients of chlorine and fluoride are calculated. At the end, simulation of part of a real network with the results reported in this study is presented.

Optimizing pump operations is a challenging task for real-time management of water distribution systems (WDS). With suitable pump scheduling, pumping costs can be significantly reduced. In this research, a novel economic model predictive control (EMPC) framework for real-time management of WDS is proposed. Optimal pump operations are selected based on predicted system behavior over a receding time horizon with the aim to minimize the total pumping energy cost. Time-varying electricity tariffs are considered while all the required water demands are satisfied. The novelty of this framework is to choose the number of pumps to operate in each pump station as decision variables in order to optimize the total pumping energy costs. By using integer programming, the proposed EMPC is applied to a benchmark case study, the Richmond Pruned network. The simulation with an EPANET hydraulic simulator is implemented. Moreover, a comparison of the results obtained using the proposed EMPC with those obtained using trigger-level control demonstrates significant economic benefits of the proposed EMPC.

The discrete choice experiment (DCE) has been widely employed to examine public preferences for changes in environmental/social welfare brought about by various policy and development proposals so as to solicit social efforts and engage civic communities to collaboratively produce sustainable societies. If DCE estimates are to inform relevant policies and decisions, it is crucial that valid responses be collected via DCE surveys. This paper examines the convergent validity of individual-level preferences for restoring an urban riverscape in Guangzhou (south China) by comparing the consistency of respondents’ stated purchasing preferences (utility) derived from the DCE approach and their theoretical stated preferences (importance) obtained from the analytic hierarchy process (AHP). Our econometric analysis of a data set, which is composed of 462 responses from an online questionnaire survey, reveals that DCE and AHP could generate similar utility-importance rankings at the group level, yet distinct discrepancies exist at the individual level. A novel finite mixture model is then recruited to explore the association of individual respondents’ utility estimate (derived from DCE) and importance weight (derived from AHP) for every attribute pertaining to the proposed river restoration, which identifies three subgroups (components) of respondents exhibiting different patterns of convergence/nonconvergence and preference heterogeneity. The AHP importance weights for respondents belonging to Component 1 (20%) and Component 2 (42%) have a statistically significant explanatory power for DCE utility estimates, suggesting that the majority of respondents provided valid responses. The respondents of Component 3 (38%) manifest a statistically insignificant relationship, showing invalid responses. Censoring this group of respondents from DCE modeling could deflate the estimates of average willingness to pay for restoring the urban riverscape and change the degree of preference heterogeneity. This pioneer individual-level validity examination of DCE sheds new light on how to combine AHP with DCE in order to derive reliable information to support decision-making.

Most urban water-use monitoring, modeling, and conservation research has focused on a large but relatively homogenous group of residential water users. Commercial, industrial, and institutional (CII) facilities use large volumes of water, but their diversity in amounts, timing, locations, and other use factors makes them difficult to monitor and study. We monitored water use at four manufacturing and two assisted-care facilities in Logan, Utah, at 5-min and 5-s frequencies for up to 1 year. We used the data to disaggregate and quantify the temporal signature of end uses that included toilets, showers, landscape irrigation, industrial processes, and humidifiers. We also quantified variability in volume, duration, frequency, and timing of end uses, variability across facilities, and average water use per end use. Results identified water-use events both specific to CII users and in common with residential users. We shared findings with participating business representatives, who verified that results matched their understanding of water-use behaviors within their facilities.

Determining beneficial operation rules for complex reservoir cascades is challenging due to the variability of river inflows, multiple water users, and the computational burden imposed by optimization calculations. Water managers of Sri Lanka face the challenge of balancing water releases for hydropower and for irrigated agriculture in the Mahaweli project. This study derived Mahaweli reservoir operation policies—reservoir rule curves and water allocation rules—by optimizing hydropower generation and agricultural yield. We derived operation rules for three reservoirs and one water diversion structure, because these offer the main opportunities for controllability of hydropower and irrigation. We used a multiobjective evolutionary algorithm to optimize hydropower and agriculture yields for 1,000 synthetically generated reservoir inflows that capture the natural variability of the monsoon seasons. The optimization problem was simplified by focusing on the main components of the cascade and by carrying out the computation in two stages. Despite the simplification, the operation rules derived using multiobjective optimization enhanced hydropower and agriculture yield compared with the values from current rules. The solution set exposed the trade-off between energy and agricultural yield objectives. Risk-neutral and risk-averse optimized rules and trade-offs between the objectives provide information that can be used to adapt water management practices for different hydrological conditions, which can be beneficial to thousands of people. Water managers of similar reservoir cascades can use the method to select policies considering trade-offs among objectives.

The Fletcher-Ponnambalam (FP) reservoir optimization method has been successfully applied to several case studies in the literature. As an explicit stochastic optimization problem, it requires the assumption of a probability distribution for the reservoir inflows in order to solve several integrals regarding the statistical moment equations for storage as well as probabilities of containment (storage within minimum and maximum levels), deficit (below minimum), and spill (above maximum). This technical note proposes an alternative way to approximate the solution of such integrals using the sample of historical inflow records. This does not involve the consideration of any distribution and leads to far shorter expressions as well as to less coding effort. The new formulation is applied to solve a water reservoir operation optimization problem for a major dam in Brazil. The results compare well with those from the current FP formulation and with those obtained by a Monte Carlo simulation of the monthly reservoir operation.

This study investigates the effect of applying scenario-based phase designs for isolation valve installation by developing an optimal multiphase valve design model while previous research focused on the single-phase design approach. During the design period, scenarios such as demand change, pipe breakage, and pipe age were considered. The model was applied to a medium-sized water distribution network, and optimal valve locations and installation timing for both single-phase and multiphase designs were determined. For the single-phase design, all valve locations were determined by forecasting changes in the future, while for the multiphase design, valves are added at the start of each phase. The results show that the single-phase design approach is more reliable and has fewer shortages than the multiphase design approach; however, the latter enables the determination of the valve installation locations based on need (e.g., the demand and changes in the pipe roughness and breakage probability). Further, cost and energy analyses show that the multiphase approach provides benefits such as low maintenance/operation costs.