Journal of Water Resources Planning and Management

The optimization of pump operations has the potential to reduce operational costs, while still maintaining the high level of reliability required of water distribution systems. The hydraulic software EPANET2 toolkit has been frequently linked to evolutionary algorithms for this purpose; however, only time-based controls and simple controls based on one single condition, e.g., the tank level, could be automatically changed during the optimization. This paper introduces a modification to the original EPANET2 toolkit library, so that the operation of pumps can be optimized taking into account simultaneously several conditions (e.g., the time of the day and the tank level). A problem in the original toolkit associated with computing pump energy and costs using rule-based controls has also been solved. The new ETTAR toolkit has been tested on a case study, in which a genetic algorithm has been used to optimize different types of controls. Results show that it is possible to find more cost-effective solutions compared to simple controls and, although with longer computational times, let the algorithm create the entire pump control rule set. The robustness of the optimized controls found has also been tested.

With rapid economic and social development, increasing water demands have made water scarcity a worldwide issue. Equitable, stable, and efficient water allocation, which includes water resources management and conflicting water demand management, needs a valid resolution. This paper develops a multiobjective bilevel model with the upper level reflecting the river basin authority’s allocation principle of equity and stability and the lower level aimed at ensuring maximal economic benefit effiency for each subarea. The Gini coefficient is introduced to seek equity in the water allocation strategies, a power index is applied to ensure stability in the water allocation solutions, and an economic efficiency function is designed to maximize the economic benefit efficiency in each subarea. The proposed multiobjective bilevel water allocation model is then applied to a case study in Qujiang River Basin in China, and demonstrates that the proposed model provides reasonable and practical water allocation strategies for river basin authorities.

The aim of this article is to develop a hybrid nested particle swarm optimization to solve a Pigovian tax-based waste load allocation problem for river systems. The river system at the Tuojiang River basin is the prototype which is then extended to a generalized waste load allocation problem. The responsible environmental protection agency (EPA), as the leader, sets the pollution tax standards at a given checkpoint to resolve the conflict between the dischargers, and each discharger, as the follower, makes biological oxygen demand (BOD) removal decisions to minimize their own pollution costs under the specified pollution and pollution tax standards. A cooperative bilevel multifollower decision-making model is established that takes into account the objectives and constraints. The particular nature of this model requires the development of a nested particle swarm optimization algorithm. Instead of using a traditional particle performance measurement method, an exact algorithm for solving the lower-level model, called a multi-agent-based dynamic extremal value algorithm (ma-DEV), is proposed and nested to deal with the bilevel model’s specific decision rule. Results for the Deyang section of the Tuojiang River are presented to demonstrate the performance of the proposed optimization method, which proved to be very effective and efficient compared to other heuristic algorithms.

In recent years, the opportunity of coupling a pressure management strategy with power generation in water distribution networks (WDNs) has attracted increasing interest. Numerical simulations developed for a number of real cases showed attractive profits and capital payback periods. Although many micro- and pico-hydro plants currently are operating along supply water mains, installations in WDNs are quite uncommon because of the high variability of flow and pressure patterns. Aiming at robust and effective industrial-based solutions, a laboratory prototype was developed for hydropower generation in WDNs. A pump as turbine (PAT) was used for generation, although a traditional turbine could be used as well. A control strategy was developed and implemented for automatic setting of regulating valves, aimed at both ensuring adequate pressure over the network and maximizing electric power. Experiments showed that the valves, equipped with the automatic control system previously described, can automatically both regulate the flow running the PAT and guarantee the set point pressure at the network monitored node, even in cases of abrupt change in pressure and flow discharge.

Best management practice (BMP) selection typically focuses on economic costs and management criteria. Generally, there is very little consideration of lifecycle assessment (LCA)-determined environmental impacts (LCA impacts) associated with management practice implementation and operation, although the inclusion of LCA considerations are becoming increasingly common within state governments and industries as part of sustainability initiatives. LCA modeling provides an evaluation of the environmental impacts of a product or process in a cradle to grave scenario following ISO 140440 protocol. LCA can provide value in preliminary decision making for construction of sustainable stormwater management structures. To assist in that effort, the LCA impacts of typical upper Midwest U.S. urban stormwater management scenarios were evaluated. Primary, traditional structures (porous detention, sand filtration basins) were considered both as stand-alone systems and as systems coupled with green BMPs (rain gardens, vegetated swales, porous pavement). Designs were based on a Rapid City, South Dakota U.S. drainage basin configuration and national stormwater design manuals. Both midpoint and endpoint LCA impacts were determined based on a 30-year design life. The functional unit used was the volume of stormwater treated by the system for a design storm of 51 mm of rain in 24 h. LCA modeling results demonstrated that although implementation of green stormwater BMP offsets did effectively reduce LCA impacts compared with traditional management strategies, there was little difference between each of the green BMPs implemented. Transportation of management practice infrastructure was the greatest LCA-impact contributor, whereas the use of locally-sourced materials provided significant benefits to the impact categories examined.

A number of multiobjective evolutionary algorithms (MOEAs) have been developed and applied to water resource optimization problems over the past decade. The comparative performance of these MOEAs has been investigated often according to their overall end-of-run results (quality of optimal fronts) within prespecified computational budgets. Despite the importance of such comparative analyses, these studies have provided little knowledge of how different MOEAs navigate through the decision space toward the Pareto front. To address this issue, this paper uses a range of metrics to quantitively characterize MOEAs’ run-time searching behavior, with a focus on the statistics of search quality and convergence progress. The metrics are applied to three state-of-the-art MOEAs, including the nondominated sorting genetic algorithm-II (NSGA-II), self-adaptive multiobjective differential evolution (SAMODE), and Borg, for six water distribution system (WDS) design problems with the objectives of minimizing network cost and maximizing network resilience. Moreover, the relationship between algorithm operators and behavioral properties is discussed. The run-time behavioral results successfully capture the underlying search characteristics associated with each MOEA, thereby offering a significantly improved understanding of how different operators affect the MOEA’s searching behavior. This insight not only offers guidance for practitioners to select appropriate MOEAs (and operators) for given optimization problems, but also builds fundamental understanding of MOEA operators, which can be used to develop further advanced optimization algorithms.

Low-impact development (LID) comprises a broad spectrum of stormwater management technologies for mitigating the impacts of urbanization on hydrological processes. Among these technologies, green roofs are one of the most adopted solutions, especially in densely populated metropolitan areas, where roofs take up a significant portion of the impervious surfaces and land areas are scarce. While the in situ hydrological performance of green roofs—i.e., reduction of runoff volume and peak discharge—is well addressed in literature, less is known about their impact on stormwater management and reuse activities at a catchment or city scale. This study developed an integrated urban water cycle model (IUWCM) to quantitatively assess the effect of uniform green roof deployment (i.e., 25, 50, and 100% conversion of traditional roofs) over the period 2009–2011 in the Marina Reservoir catchment, a $100\text{-}{\mathrm{km}}^2$, highly urbanized area located in the heart of Singapore. The IUWCM consists of two components: (1) a physically based model for extensive green roofs integrated within a one-dimensional numerical hydrological-hydraulic catchment model linked with (2) an optimization-based model describing the operation of the downstream, stormwater-fed reservoir. The event-based hydrological performance of green roofs varied significantly throughout the simulation period with a median of about 5% and 12% for the catchment scale reduction of runoff volume and peak discharge (100% conversion of traditional roofs). The high variability and lower performance (with respect to temperate climates) are strongly related to the tropical weather and climatic conditions—e.g., antecedent dry weather period and maximum rainfall intensity. Average annual volume reductions were 0.6, 1.2, and 2.4% for the 25, 50, and 100% green roof scenarios, respectively. The reduction of the stormwater generated at the catchment level through green roof implementation had a positive impact on flood protection along Marina Reservoir shores and the energy costs encountered when operating the reservoir. Vice versa, the drinking water supply, which depends on the amount of available stormwater, decreased due to the evapotranspiration losses from green roofs. Better performance in terms of stormwater reuse could only be obtained by increasing the time of concentration of the catchment. This may be achieved through the combination of green roofs with other LID structures.