Journal of Water Resources Planning and Management

This paper presents an efficient and reliable swarm intelligence-based approach, namely elitist-mutated particle swarm optimization (EMPSO) technique, to derive reservoir operation policies for multipurpose reservoir systems. Particle swarm optimizers are inherently distributed algorithms, in which the solution for a problem emerges from the interactions between many simple individuals called particles. In this study the standard particle swarm optimization (PSO) algorithm is further improved by incorporating a new strategic mechanism called elitist-mutation to improve its performance. The proposed approach is first tested on a hypothetical multireservoir system, used by earlier researchers. EMPSO showed promising results, when compared with other techniques. To show practical utility, EMPSO is then applied to a realistic case study, the Bhadra reservoir system in India, which serves multiple purposes, namely irrigation and hydropower generation. To handle multiple objectives of the problem, a weighted approach is adopted. The results obtained demonstrate that EMPSO is consistently performing better than the standard PSO and genetic algorithm techniques. It is seen that EMPSO is yielding better quality solutions with less number of function evaluations.

The direct search approach to determine optimal reservoir operating policies is proposed with a real coded genetic algorithm (GA) as the optimization method. The parameters of the policies are optimized using the objective values obtained from system simulations. Different reservoir release rules or forms, such as linear, piecewise linear, fuzzy rule base, and neural network, are applied to a single reservoir system and compared with conventional models such as stochastic dynamic programming and dynamic programming and regression. The results of historical and artificial time series simulations show that the GA models are generally superior in identifying better expected system performance. Parsimony of policy parameters is inferred as a principle for selecting the structure of the policy, and Fourier series can be helpful for reducing the number of parameters by defining the time variations of coefficients. The proposed method has shown to be flexible and robust in optimizing various types of policies, even in models that include nonlinear, nonseparable objective functions and constraints.

Many water authorities, both nationally and internationally, have been forced to rethink their strategies for achieving water balance as a result of growing water demands, droughts, reduced no-failure yields, and environmental sustainability considerations. In particular, regulatory bodies in Australia are demanding that water managers exhaust network management efficiencies before considering new water source options (e.g., dams, desalination, pipelines, etc.). Demand management incentive schemes in conjunction with water recycling and pressure and leakage management (PLM) initiatives are a few examples of least-cost planning strategies being adopted by water authorities to achieve water balance without expanding the water infrastructure asset requirements. Potential benefits of PLM strategies have been predicted by these authorities worldwide in areas such as deferred capital works, reduced corrective maintenance, reduced treatment costs, energy savings, reduced reclaimed water discharges, and improvements to customer service. However, justification for PLM options remains difficult due to the limited amount of quantified evidence for most of the above-mentioned benefits over an urban water system’s life cycle. As the first stage in the development of a holistic PLM decision support system, this paper quantifies the benefits derived from a PLM strategy in a trial area located on the Gold Coast, in Queensland, Australia. The results of the trial provide evidence to support claims that PLM can reduce water consumption and the frequency of infrastructure failures if implemented throughout the entire Gold Coast City. Furthermore, the research concludes that PLM impacts on the total water cycle and has broad implications for ensuring the future sustainability of potable water services.

Six storm-water best management practices (BMPs) for treating urban rainwater runoff were evaluated for cost and effectiveness in removing suspended sediments and total phosphorus. Construction and annual operating and maintenance (O and M) cost data were collected and analyzed for dry extended detention basins, wet basins, sand filters, constructed wetlands, bioretention filters, and infiltration trenches using literature that reported on existing storm-water BMP sites across the United States. After statistical analysis on historical values of inflation and bond yields, the annual O and M costs were converted to a present worth based on a 20-year life and added to the construction cost. The total present cost of each storm-water BMP with the 67% confidence interval was reported as a function of the water quality design volume, again with a 67% confidence interval. Finally, the mass of total suspended solids and total phosphorus removed over the 20-year life was estimated as a function of the water quality volume. For the six storm-water BMPs investigated, results show that, ignoring land costs, constructed wetlands have been the least expensive to construct and maintain if appropriate land is available. However, since wetlands typically require more land area to be effective, land acquisition costs may result in wetlands being significantly more expensive than other storm-water BMPs that require less area. The results can be used by planners and designers to estimate both the total cost of installing a storm-water BMP and the corresponding total suspended solids and total phosphorus removal.

An artificial neural network methodology is developed to investigate the possible effects on monthly and seasonal surface water supplies in Colorado’s Arkansas River Basin under two transient climate change scenarios, the HAD from the Hadley Centre for Climate Prediction and Research and the CCC from the Canadian Climate Centre. The results show that the decade-to-decade variability is considerably more apparent than any long-term trend or change. Under the HAD scenario, water available for irrigation is expected to increase above the historical baseline in every month of the growing season. However, the CCC scenario predicts constant water shortages in the region and decreased water available for irrigation in almost every month. This wide variation in the predictions from the HAD and CCC scenarios means that there is a large degree of uncertainty on what the future impacts of climate change might be in the region. However, the methodology developed can be used to estimate the impacts of new or updated predictions of climate change.

This paper presents a stochastic design approach that quantifies the expected annual damages sustained by residential, commercial, and industrial users during low- and high-pressure hydraulic failures in a water network. The approach, which couples stochastic models of water demand, fire flow, and pipe breaks with Monte Carlo simulation, was used to solve part of the Anytown design problem. Results indicated that a significant proportion of low-pressure failures occurred during low-demand months in the last $10\text{years}$ of the planning period. The timing and spatial distribution of demands observed during failure differed significantly from the demands assumed in conventional design (maximum hour and maximum day $\text{demand}+\text{fire})$ . The results also indicated that including damages in design makes it possible to produce cost effective systems that yield a low level of expected annual damages. This gives force to framing the network design problem as a stochastic, multiobjective one to balance cost efficiency with system capacity/redundancy and provide a hedge against hard-to-anticipate temporal and spatial patterns of demand in networks.

Many groundwater remediation designs for contaminant plume containment are developed using mathematically based groundwater flow models. These mathematical models are most effective as predictive tools when the parameters that govern groundwater flow are known with a high degree of certainty. The hydraulic conductivity of an aquifer, however, is uncertain, and so remediation designs developed using models employing one realization of the hydraulic conductivity field have an associated risk of failure of plume containment. To account for model uncertainty attributable to hydraulic conductivity in determining an optimal groundwater remediation design for plume containment, a method of optimization called robust optimization is utilized. This method of optimization is a multiscenario approach whereby multiple hydraulic conductivity fields are examined simultaneously. By examining these fields simultaneously, the variability of the uncertainty is included in the model. To increase the efficiency of the robust optimization approach, a sampling technique is developed that allows the modeler to determine the minimum number of field realizations necessary to achieve a reliable remediation design.

The allocation of irrigation water to multiple, spatially distinct sites is one of the oldest and most important problems in water resources planning and management. However, past technical investigations have focused almost exclusively on allocation for an optimization objective and have not diagnosed the tradeoffs existing between this objective and other important socioeconomic goals such as equity and security. A common optimization technique is reviewed and adapted to multiple levels of irrigation allocation. The process of spatial distribution is then investigated for five different hypothetical allocation scenarios, which represent various definitions of the objectives of efficiency, equity, and security. A semihypothetical case study for the Lake Victoria Basin in East Africa is used to illustrate international water and crop production distributional effects of the five water-sharing scenarios. It is found that pursuit of nonefficiency objectives does entail material costs. The magnitude of these costs is variable with the specific policy pursued and the total system allocation level. Pursuit of a single objective, equity, can lead to very different patterns of water allocation and crop production depending on how exactly the political-social idea is defined. Finally, physical limitations of the agricultural system may require an explicit choice between objectives as it is shown that food supply security may not be possible under certain equity-targeted policies.

Although agriculture and its use of water resources are obviously sensitive to climatic conditions, past research has seldom identified the effects of climate variability and climate change on the fully developed relationship between crop yield and irrigation. There is potentially great value in understanding the role of climatic uncertainty on this relationship because of the dependence of agriculture on irrigation and the scale of water consumption for irrigation. Probability distributions of crop-water production functions (CWPF-PDs) are demonstrated as being useful encapsulations of the climate-yield-irrigation relationship for decisions at various levels of time and space. Combined with reliable climate teleconnections or climate forecasts, CWPF-PDs can be a central decision support tool for questions of risk and reliability. For long planning horizons, potential climate change predicted by multiple general circulation models (GCMs) can be assessed in the context of agricultural water resources. By analysis of changes in the CWPF-PDs, conclusions regarding the efficacy and sustainability of water resources and agricultural policies can be made. A semihypothetical case study for the Lake Victoria Basin in East Africa is used to illustrate these methodologies, and potential future climate impacts as predicted by the CGCM1, ECHAM3, and HadCM2 GCMs are discussed.