Journal of Water Resources Planning and Management

There are multiple sources of uncertainties in urban water-distribution systems, e.g., nodal water demand and sensor measurement error. All of these uncertainties increase the complexity of contaminant source identification in a sparse sensor network. The large number of attributes (e.g., contaminant source location, magnitude, injection starting time, and duration) of a contaminant event profile cannot be identified given limited sensor data. Instead, the uncertainties in the contaminant event profile need to be characterized. Markov chain Monte Carlo (MCMC) methods for Bayesian analyses allow for the characterization of the uncertainty in the contamination event profile. To account for stochastic water demands, which has been shown in some circumstances to be necessary if the contaminant event is to be properly characterized, the evaluation of the likelihood function is the most computationally expensive part of the MCMC implementation. Previous work applied Monte Carlo methods for error propagation (MCEP). The research reported in this paper investigates the application of support vector regression (SVR) to speed the likelihood evaluation. This coupled MCMC-SVR approach enables probabilistic inference of the contaminant event. An SVR model, which maps from the contaminant event space to the likelihood space, is built for each node in the network to evaluate the likelihood function during the MCMC chain evolution. For the case study investigated, MCMC-SVR is computationally feasible and robust in inferring the contaminant event. A comparison between MCMC-SVR and MCMC-MCEP reveals that there is no substantial difference between the inferences provided by the two models, whereas the former is more computationally efficient.

Once contamination of a drinking water distribution system is suspected or known, emergency managers may take different actions in response to the perceived state of the system. This study explores performance of a novel protective response action—injection of food-grade dye directly into drinking water—for minimization of health impacts as a contamination threat unfolds. Dye injection acts as an alerting mechanism that discourages public consumption of potentially contaminated water. Considering the uncertainties in threat observations and the imperfection in system understanding, however, the action has potential for costly false alarms. These could occur when contamination has indeed happened but population segments residing in safe regions are mistakenly alerted or when suspected indications of contamination occurrence turn out to be entirely wrong. The emergency response is, thus, formulated as a multiobjective optimization problem for the minimization of risks to life with minimum public warning and execution cost. Dye injection rates, locations, and timing are explicitly treated as optimization decision variables and sensitivity analyses are conducted on number of injection locations, response delay, and contamination scenario characteristics. Application to a virtual water distribution system demonstrates that dye injection holds potential as an effective contamination emergency response strategy. The results represented as a set of diverse nondominated alerting protocols provide valuable insight into optimal trade-off between achievement of health protection and avoidance of false alarms to public. The modeling framework can be useful for the management of contamination events for diverse types of contaminant agents without unintended damages to the system and interruption of firefighting.

In recent years, a number of evolutionary algorithms have been proposed for optimizing the design and operation of water distribution systems (WDSs). These evolutionary algorithms include genetic algorithms, ant colony optimization, particle swarm optimization, the shuffled leaping frog algorithm, and differential evolution. Although there have been some comparisons made of the performance of the various algorithms, very few of these comparisons have been carried out in a completely rigorous manner. The main aim of this paper is to introduce a methodology for the rigorous comparison of various algorithms for the optimum design of water distribution systems. The methodology involves comparing the various algorithms in terms of (1) the best solution obtained; (2) the speed of convergence; and (3) the spread and consistency of the solutions obtained over a number of random starting seeds and numbers of evaluations. As a demonstration of the methodology, the techniques of genetic algorithms (GA), particle swarm optimization (PSO), and differential evolution (DE) are applied to two frequently used WDS case studies, namely the New York Tunnels and Hanoi water networks. In addition, the techniques are applied to a real-size water distribution system consisting of 476 pipes. The results obtained show that the algorithm performances depend on the specific problem and the number of function evaluations allowed. Moreover, it is shown that correct calibration is an essential phase for a fair comparison of evolutionary algorithms. In fact, the best parameters are a function of the problem characteristics, of the objective function and of the variants in the algorithm operators. Therefore the adoption of configurations tested on slightly different versions of the algorithms can lead to quite different results.

The hydraulic reliability of bulk water-supply systems can be defined in terms of the failure frequency of the municipal storage tanks they supply: the system fails when the tank runs dry and is functional otherwise. Municipal storage tanks are normally sized according to deterministic guidelines that make allowances for balancing, fire and emergency storage. In this study, genetic algorithm (GA) optimization was used together with stochastic analyses to find the optimal combination of feeder pipe configuration, feeder pipe capacity and tank capacity for a given risk of failure. A sensitivity analysis was conducted to determine the factors that have the greatest impact on the optimal design for an example system. The results showed that the optimal pipe configuration is a single feeder pipe in most cases, but that two parallel pipes are preferable for shorter feeder pipes. It was found that it is often cost-effective to trade off a smaller tank size for a greater feeder pipe capacity. Based on this finding, design guidelines are likely to specify suboptimal solutions in many cases.

Long-range infrastructure planning is fraught with uncertainties. Population growth may occur faster or slower than expected, regulations may change, and public sentiment can shift. In the face of these uncertainties, water system managers must plan for large infrastructure investments and the questions about when it is desirable to invest in infrastructure and what is the appropriate infrastructure component size to meet growing demands. One of the most powerful and intuitive ways to incorporate uncertainties is to use scenarios that represent plausible futures. Scenario-based planning is gaining acceptance in the water resources community. Preparing for a range of possible futures provides flexibility and adds robustness to the system so it can respond to uncertain events at reasonable costs while maintaining community confidence in their utilities. In this paper, novel scenario-based planning and optimization approaches are presented for the optimal design of regional-scale water supply infrastructure in a multiperiod planning framework. For demonstration, water demand projections are considered as uncertain and multiperiod construction projects are selected to minimize the economic costs. The solution approaches are applied to a decentralized water reclamation planning project in a green-field development area in southeast Tucson, Arizona, where water reclamation is viewed as a potential component for a sustainable water supply.

In this paper, an overview of a strategy for automatic meter reading (AMR) data interpretation and aggregation is presented along with the proposed stochastic models adequate for representing the intrinsic characteristics of the data. Water demand measurements from single user accounts are obtained from an AMR system that continuously monitors consumption in different zones of Cincinnati, Ohio. The data represent volumetric measurements characterized by fixed increments, which depend on the sensitivity of the instruments used and occur at irregular times due to the polling method of the AMR system. Given the nature of the data, a nonhomogeneous Poisson process is proposed to model the arrivals of the increments within a selected time interval of 350 days. An exponential-polynomial-trigonometric rate function with multiple periodicities (EPTMP) is assumed to describe both trends and periodicities in the observed data. A specific methodology for estimating the parameters of the EPTMP rate function is presented, based on the method of maximum likelihood. In order to evaluate the estimation technique, a performance evaluation is carried out on synthetic data generated in simulation. Finally, the estimation method is applied and tested on samples of the complete AMR data set, which is obtained from aggregating randomly selected subsets of different magnitude. The results provide significant evidence of the numerical stability and accuracy of the modeling procedure and encourage the use in simulation and prediction of water demands at network nodes from available AMR data.

Pipeline leak detection using hydraulic transient analysis is a relatively new detection technique. For single pipeline systems, recent work has led to two different approaches for determining leak parameters based on leak-induced patterns displayed in a pipeline’s frequency response diagram (FRD). The major difference between the two techniques is that one uses the leak-induced pattern within the odd harmonics of an FRD, while the other one uses the leak-induced pattern at the even harmonics. In order to compare and contrast the two approaches, the current research analyses the relationship between the characteristics of the leak-induced patterns and the parameters of the pipeline system. A dimensionless analysis, based on hydraulic impedance, is adopted to simplify the equations. The amplitudes of leak-induced patterns at both the odd and even harmonics in the FRD are found to be dependent on a critical parameter: the dimensionless steady-state valve impedance, $Z_V^{*}$. The value of $Z_V^{*}$ is dependent on the steady-state valve opening. As a result, amplitudes of the leak-induced patterns within the FRD for any specific pipeline system can be controlled by the initial valve opening. This research also derives the equations for calculating the dimensionless leak size based on the value of $Z_V^{*}$ and the amplitude of either leak-induced pattern. Finally, the two existing FRD-based leak detection methods are compared, and the approach using the odd harmonics is found to be superior.

The availability of water resources in many urbanizing areas is the emergent property of the adaptive interactions among consumers, policy, and the hydrologic cycle. As water availability becomes more stressed, public officials often implement restrictions on water use, such as bans on outdoor watering. Consumers are influenced by policy and the choices of other consumers to select water-conservation technologies and practices, which aggregate as the demand on available water resources. Policy and behavior choices affect the availability of water for future use as reservoirs are depleted or filled. This research posited urban water supply as a complex adaptive system (CAS) by coupling a stochastic consumer demand model and a water supply model within an agent-based modeling (ABM) framework. Public officials were simulated as agents to choose water conservation strategies and interbasin transfer strategies, and consumers were simulated as agents, influenced by various conservation-based programs to select water conservation technologies and behaviors, and correspondingly update their individual demand models. A water supply reservoir was simulated to receive rainfall from the contributing watershed and to supply the demands of consumer agents. The ABM framework was applied to an illustrative urban case study. A set of scenarios was developed to represent moderate and strong water conservation strategies, and was simulated for a long-term precipitation record to evaluate the sustainability of water conservation practices.

This paper presents a methodology for obtaining and adjusting of efficiency curves for hydroelectric generating units. It is based on measured data of power, gross head, and water discharge recorded by the company that manages the plant operation. The objective is to determine the actual performance characteristics of the set: turbine, generator, and penstock. In order to obtain the efficiency functions, an iterative calculation is used. Its input data are the functions currently in use of turbine efficiency, generator efficiency, and penstock head losses. For the adjustment of the efficiency functions, the Generalized Reduced Gradient optimization method is employed. A case study was applied to the data from a large Brazilian hydroelectric plant whose operation is under the coordination of the Electric System National Operator. The benefits of the proposed methodology are analyzed using a simulation tool for the hydroelectric operation. The simulator is used to reproduce the past operation of the plant, first with current data and second with adjusted data. The results show that the optimal unit efficiency functions significantly contribute to bring the real and simulated operation closer.

Optimal design of water distribution networks (WDN) involves an evaluation of both consumers’ pressure benefits and investors’ economic objectives. The aforementioned objectives often conflict, so finding the optimal solution for one of those objectives reduces the other objective’s utility. In such situations, there are many nondominated solutions, each solution denoting an alternative that cannot be preferred over another in terms of both objectives. Thus, an appropriate alternative to fulfill both objectives and satisfy decision makers’ criteria and meet the design purposes within a desirable range necessitates the use of bargaining models that are called conflict-resolution models. This paper considers two urban WDN optimization design problems having different objectives, including initial costs and hydraulic performance improvement of the network by satisfying given hydraulic constraints. First, a set of alternatives are drawn out by a fast messy genetic algorithm (FMGA), then the appropriate alternative design is achieved by using Nash’s and Young’s bargaining models. Thus, the methodology presented in this paper can be employed by designers of WDNs to simultaneously consider the utilities of both consumers and investors that are the main beneficiaries of such infrastructures. Results show that in almost all the alternatives obtained by the proposed methodology, at least 80% of the most probable utility of both beneficiaries is obtained at the same time, which indirectly indicates the low vulnerability of the design alternative by the considered methodology in meeting the goals of each beneficiary. Moreover, results also indicate that when using the same utility functions, decision points obtained from the two models coincide.

Water transfers are one method of allowing utilities to meet demand during dry periods while avoiding, or at least forestalling, the construction of costly new supply capacity. Nonetheless, transfer agreements must have clearly defined terms and decision rules to be effectively implemented. In developing these rules, careful consideration should be given to the risk tolerance of both the buying and selling utilities because these factors can significantly impact the nature of the agreements. This study uses a simulation approach to evaluate interutility transfer agreements that include different mechanisms to reduce risk for both the buyer and seller. For the seller these can include seasonal and volume-based transfer limits that ensure its ability to meet the demands of its own customers before making transfers. For the buyer, important features involve defining the conditions under which it can request transfers, a choice dictated by the buyer’s risk tolerance. Several potential agreement structures are considered, with the volume and frequency of transfers as well as the costs varying considerably depending on the risk-reduction mechanisms incorporated. Results indicate that more risk-averse contract structures can significantly increase costs relative to more risk-tolerant scenarios, even though the same reliability objectives are met. While some level of risk aversion is warranted based on uncertainties related to factors such as future demand growth and climate change, the degree of risk aversion justified is a question that deserves greater scrutiny in water resource management plans. In this case, even the most risk-averse agreements were still less expensive than comparable structural alternatives for improving supply reliability.

Excessive nitrate loads in surface waters are a major cause of hypoxia and eutrophication. In many places, agriculture is the single largest source of nitrogen entering receiving waters. Perennial energy grass crops have the potential to reduce nitrogen loads from agricultural areas, while sequestering carbon and offering new economic opportunities for farmers. This study analyzes farm system-scale cropping and fertilizer application decisions, and resulting nitrate loads, as driven by prices for the bioenergy crop miscanthus, as well as investigates reductions of carbon and other greenhouse gas emissions and nitrogen fertilizer use. An economic model of farm-system-scale decisions is coupled to a hydrologic-agronomic model of the physical stream system to obtain nitrate loading and crop yield results for varying combinations of prices and policies for a typical Midwestern agricultural watershed. For the scenarios examined, a large reduction in stream nitrate load depends on a high price for miscanthus relative to competing crops. A price for miscanthus that exceeds 50% of the average of corn and soybean prices, per unit weight, is estimated to lead to nitrate load reductions of 25% or more. Though significant, these reductions are still less than the recommended 45% reduction in stream nitrogen flux entering the Gulf of Mexico needed to mitigate the hypoxia problem in the gulf. Miscanthus prices are unlikely ever to reach such levels. However, nitrate load reductions could still be achieved by implementing a nitrogen fertilizer reduction subsidy alongside a miscanthus market. The results also show that carbon trading is unlikely to result in any significant reduction in nitrate load. The results are useful for improving understanding of the potential of these incentives, individually and concurrently, to reduce pollution from Midwestern crop agriculture.

This paper presents a data-driven simulation-based burst detection approach in water distribution systems using control limit analysis. System specific burst sensitivity tables are developed by synthetically generating burst events and analyzing the system hydraulic responses to the given bursts. Potential locations of pressure and flow meters are suggested based on the sensitivity tables. By matching the field observation data with the developed burst sensitivity tables, potential burst hotspots are identified. Using simulated bursts for a simple literature network, it was found that the proposed approach is effective to quickly locate bursts and reduce response times.