Journal of Water Resources Planning and Management

This paper addresses a knowledge-based approach for reservoir system optimization. The approach takes a more detailed consideration of each turbine in a hydropower plant than the traditional constant output coefficient method. To use this approach, a knowledge expression and a knowledge function are defined for hydropower plant operation. The knowledge expression is extracted by dynamic programming to save all possibly optimal situations of unit commitment, and further, the knowledge function is formulated based on a two-dimensional interpolation of the knowledge expression. Through the use of the knowledge expression and the knowledge function, computer memory requirements can be reduced and unnecessary computations can be avoided in the reservoir operation optimization. To overcome the decomposition schemes in time and in space and guarantee finding the global optimum (in a discrete sense) with an extended time horizon, up to 400 CPU cores are used to run a parallel dynamic programming model, which applies the knowledge-based approach, to estimate the maximum energy production of the Three Gorges Project (TGP) and the Gezhouba Project (GZB) cascade hydropower plants in China in the year of 2010, with 1 day as the time step and 365 days as the time horizon. The case study results show that the maximum energy production of the TGP-GZB system would be $\mathrm{1,035.0}\times {10}^8\mathrm{kWh}$ in 2010 under the current operating rules. Thus, there is still room for improvement in energy production, with the maximum increase of 1.83% ($18.3\times {10}^8\mathrm{kWh}$) from optimizing hydropower plant operation and 3.46% ($34.6\times {10}^8\mathrm{kWh}$) from optimizing both hydropower plant operation and reservoir operation. The overall approach is effective in providing optimal assessment solutions with an acceptable computation time. Thus, it can be concluded that the approach can be applied for planning purposes or providing more reasonable boundaries for real-time operation.

The use of microhydroelectric plants in urban pipe networks, based on the combination of a pump as turbine (PAT), two regulating valves, and two pressure meters, is proposed along with simple automation rules. Its economic benefit is tested on a small pipe network, where the network geometry as well as the demand coefficient variation in time and space have been inferred from previously collected data and existing analysis. A similar analysis has been also carried out for different scenarios in which the reduction of pipe installation cost due to a diameter reduction is compared with the increased benefit in energy production. The case study shows that a small increment of the pipe installation cost, with respect to the minimum required by the nodes minimum pressure, can lead to a larger benefit for energy production.

A valve-enclosed segment is the basic unit of water distribution networks (WDNs) in reliability analysis and optimal design. Based on the theory of the transitive closure of graphs, this paper presents a new method for identifying the whole segments in large WDNs in a single run. The method clones real valves as fictitious valves, uses valve pairs to simulate valve on-off status, and improves the Warshall algorithm to compute node transitive closure sets to identify all segments at a time. Additionally, redundant valves, endpoint valves, leaf segments, orphan segments, and regular segments are analyzed. One-way valves are considered in an unintended segment analysis. The method is described and discussed in detail by using a small sample network, then applied to a large real network containing 69,815 valves to prove its capability and efficiency.

Municipalities with limited budgets will have to rehabilitate their stock of deteriorated water mains in the coming decades. Optimization and asset management offer municipalities the opportunity to plan the rehabilitation of their water mains in a more rational manner. The aim of this paper is to develop a new event-based approach to optimize the timing of water main rehabilitation. The approach incorporates a new gene-coding scheme and covers the full range of decisions about pipe replacement, duplication, lining, new pipe installation, and asset management strategies (infrastructure adjacency and quantity discounts applied to the installed pipe). The new approach is applied to the Fairfield water network, and results suggest that applying a budget constraint prohibits early investment in pipe rehabilitation with a resulting increase in leakage, pipe breaks, and energy costs. Applying pipe discounts decreases capital and operation costs and favors pipe lining over pipe replacement and duplication in the Fairfield network. A sensitivity analysis suggests that uncertainties in demand, leakage, and break growth rate have a moderate to significant impact on capital and operation costs.

Hydraulic transients pose many risks to urban water distribution systems. The high pressure induced by a hydraulic transient can cause pipe bursts, the negative pressure can cause pipe collapses, cavitation can cause pipeline erosions and pump impeller damages, and the transient forces can loosen pipe joints. We derived the hydraulic calculating equations based on the method of characteristics (MOC) by considering the pipe head loss and node cavitation. Four individual factors and one composite factor for hydraulic transient risk are presented along with the method of calculation of the indicators. They can produce a tabular result that is color coded by risk, and the risk classification map can be built by a four-layer state assessment and color coding technology. They are Class I with green for safe level, Class II with cyan for warning level, Class III with blue for dangerous level, and Class IV with red for severe level. The pipe burst risk assessment methods and procedures are presented. Finally, a sample model and a Chinese city pipe network system are presented for pipe burst risk classification examples.

Much progress has been made in the standardization of uncertainty analysis techniques for simulation modeling but less progress has been made in optimization modeling. Among the various techniques used for optimization modeling under uncertainty, robust optimization (RO) uniquely allows for evaluation and control of the various risks of poor system performance resulting from input parameter uncertainties in water-resources problems. A model formulation was developed that addresses an inadequacy in a previous RO formulation. The importance of evaluating, through postprocessing, RO model results with respect to a range of performance metrics, has been demonstrated rather than a single metric, as has been common in previous studies. An analysis of the tradeoffs between solution robustness (nearness to optimality across all scenarios) and feasibility robustness (nearness to feasibility across all scenarios) illustrates the importance of including these terms in multiobjective water resources decision models.

Physically based semidistributed rainfall-runoff models are an important tool for assessing decentralized green infrastructure alternatives in controlling combined sewer overflows (CSO). Few studies have analyzed reliable calibration methods of highly detailed rainfall-runoff models at the subcatchment level. This research presents calibration of the storm water management model version 5 (SWMM5) using high temporal resolution rainfall and flow data to perform a 10-month continuous simulation on a physically distributed urban catchment. Detailed analyses of parameters that physically represent the area in the SWMM5 model were done using remote sensing and geographical information systems techniques. Parallel computing and a multi-search driver were implemented along with a model-independent parameter estimation (PEST) method, to find global optima and accelerate the calibration process. PEST internal settings were tuned for the continuous SWMM5 estimation problem using one month of rainfall data. Subsequently, parameter calibration was carried out with 10 months of rainfall data and model validation was performed with 5 months of data. The overall calibration and validation of the SWMM5 model was good with a Nash-Sutcliffe efficiency coefficient greater than 0.5 and a total volume runoff error of less than 5%. Time offset bias from the rainfall data affected negatively the Nash-Sutcliffe efficiency because of highly variable storm cells. This study shows it is possible to reproduce accurately the urban runoff from a highly detailed, semidistributed SWMM5 model, even when the calibration process is restricted to physical parameters and not allowed to distort the geometric representation of the project area.

In recent research, the residual of the Kalman filter applied to flow measurements in water distribution systems has been found to strongly correlate with bursts. A positive residual represents an unexpected extra flow, which is likely caused by an event such as a burst in the downstream network. However, there is a certain level of uncertainty in the flow measurements, leading to a noisy and fluctuating residual during normal and abnormal network operations. This work investigates how changes to a residual threshold affect the success rates of burst alarming during burst and nonburst periods, respectively, and presents a statistical method to automatically select a suitable residual threshold to tolerate uncertainty in flow measurements. In addition, this work also investigates how the number of burst alarms is influenced by using different flow measurement sampling frequencies and different averaging window sizes in the data sets. Engineered tests with three simulated burst events were conducted to validate the methods. The results showed that the threshold proposed in this study can produce a relatively high success rate in burst alarming. In addition, the results also showed that the current sampling interval of 15 min is suitable for burst detection from flow data, and that the use of a window averaged over the past several hours can reduce continual on/off alarm states.