Search

In this study, a model predictive control (MPC) system is developed for the goldfield and agricultural water system (GAWS) east of Perth in Western Australia. As part of the study, four months' water quality and hydraulic data of the system were collected for the development of the MPC system. Two artificial neural network (ANN) models are developed to model the relationships between the control variable, the ammonia dosing rate at the source, and the controlled variables, the total chlorine and free ammonia levels at a designated location (Goomalling pump station) in the network five days later. A two-step process based on both mutual information (MI) and partial mutual information (PMI) is used to select appropriate inputs for the total chlorine and free ammonia models. The total chlorine and free ammonia ANN models perform well, with validation Nash-Sutcliffe efficiencies of 0.84 and 0.62, respectively, and validation root mean square errors (RMSE) of 0.1320 and 0.0106 mg=L, respectively. A real-number coded genetic algorithm is then used to find the optimal ammonia dosing rate to achieve the target total chlorine and free ammonia levels at the modeled location. The results demonstrate that the developed MPC system can control the total chlorine and free ammonia levels at Goomalling pump station to be close to their target values by adjusting the ammonia dosing rates at Mundaring pump stations. The errors in the MPC system are mainly due to the relatively weak relationship between the control and controlled variables for this particular system. [ABSTRACT FROM AUTHOR]

Human water use is increasing and, as such, water for the environment is limited and needs to be managed efficiently. One method for achieving this is the scheduling of environmental flow management alternatives (EFMAs) (e.g., releases, wetland regulators), with these schedules generally developed over a number of years. However, the availability of environmental water changes annually as a result of natural variability (e.g., drought, wet years). To incorporate this variation and schedule EFMAs in a operational setting, a previously formulated multiobjective optimization approach for EFMA schedule development used for long-term planning has been modified and incorporated into an adaptive framework. As part of this approach, optimal schedules are updated at regular intervals during the planning horizon based on environmental water allocation forecasts, which are obtained using artificial neural networks. In addition, the changes between current and updated schedules can be minimized to reduce any disruptions to long-term planning. The utility of the approach is assessed by applying it to an 89km section of the River Murray in South Australia. Results indicate that the approach is beneficial under a range of hydrological conditions and an improved ecological response is obtained in a operational setting compared with previous long-term approaches. Also, it successfully produces trade-offs between the number of disruptions to schedules and the ecological response, with results suggesting that ecological response increases with minimal alterations required to existing schedules. Overall, the results indicate that the information obtained using the proposed approach potentially aides managers in the efficient management of environmental water. [ABSTRACT FROM AUTHOR]

Optimization of large and hydraulically complex water distribution systems (WDSs) is computationally expensive as simulation models are required to evaluate the performance of solutions to the problem at hand. Metamodels can act as a surrogate or substitute for these simulation models and provide significant speed-ups in the optimization process. The application of metamodels in the field of WDS optimization has been limited to date, and little guidance has been given in terms of constructing metamodels for hydraulically complex systems. While it is relatively straightforward to obtain satisfactory metamodel approximations to simulation models of simple WDSs, this is not necessarily the case for more complex networks. In order to reduce the complexity of the relationship that is to be approximated by the metamodels, a number of factors have to be considered, including the complexity of the hydraulic simulation model, the complexity of the decision space, and the locations at which outputs are required from the hydraulic simulation model. This research presents a systematic methodology for dealing with these factors and demonstrates the effectiveness of the approach by applying it to an actual WDS. A system in Wallan, Victoria, Australia, is selected for demonstration purposes. Four different metamodelling scenarios are presented here. The results show that, for this case study, some skeletonization of the model is required to achieve suitably accurate metamodels. The optimization results show a reduction in the average daily pumping costs from $457 to $363; a saving of 21%. The net present value (NPV) over 25 years is used as the objective function, which includes both pumping and chlorine costs. The current operating regime corresponds to an NPV of $1.56 million, while the optimized solution has an NPV of $1.34 million; a saving of 14%. In addition to these economic benefits, the optimized solution achieves adequate disinfection throughout the system, whereas the current operating regime results in deficits in chlorine residuals at several locations in the system. [ABSTRACT FROM AUTHOR]

Genetic algorithms (GAs) have been shown to apply well to optimizing the design and operations of water distribution systems (WDSs). The objective has usually been to minimize cost, subject to hydraulic constraints such as satisfying minimum pressure. More recently, the focus of optimization has expanded to include water quality concerns. This added complexity significantly increases the computational requirements of optimization. Considerable savings in computer time can be achieved by using a technique known as metamodeling. A metamodel is a surrogate or substitute for a complex simulation model. This research uses a metamodeling approach to optimize a water distribution design problem that includes water quality. The type of metamodels used are artificial neural networks (ANNs), as they are capable of approximating the nonlinear functions that govern flow and chlorine decay in a WDS. The ANNs were calibrated to provide a good approximation to the simulation model. In addition, two techniques are presented to improve the ability of metamodels to find the same optimal solution as the simulation model. Large savings in computer time occurred from training the ANNs to approximate chlorine concentrations (approximately 700 times faster than the simulation model) while still finding the optimal solution. [ABSTRACT FROM AUTHOR]

Copyright of Canadian Geotechnical Journal is the property of Canadian Science Publishing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In recent years, artificial neural networks have been used successfully to model multivariate water resources time series. By using analytical approaches to determine appropriate model inputs, network size and training time can be reduced. In this paper, it is proposed that the method of Haugh and Box and a new neural network-based approach can be used to identify the inputs for multivariate artificial neural network models. Both methods were used to obtain the inputs for a multivariate artificial neural network model used for forecasting salinity in the River Murray at Murray Bridge, South Australia. The methods were compared with a third method that uses knowledge of travel times in the river to identify a reasonable set of inputs. The results obtained indicate that all three methods are suitable for determining the inputs for multivariate time series models. However, the neural network-based method is preferable because it is quicker and simpler to use. Any prior knowledge of the underlying processes should be used in conjunction with the neural network method. [ABSTRACT FROM AUTHOR]