Journal of Water Resources Planning and Management

With construction of new large-scale water storage projects on the wane in the U.S. and other developed countries, attention must focus on improving the operational effectiveness and efficiency of existing reservoir systems for maximizing the beneficial uses of these projects. Optimal coordination of the many facets of reservoir systems requires the assistance of computer modeling tools to provide information for rational management and operational decisions. The purpose of this review is to assess the state-of-the-art in optimization of reservoir system management and operations and consider future directions for additional research and application. Optimization methods designed to prevail over the high-dimensional, dynamic, nonlinear, and stochastic characteristics of reservoir systems are scrutinized, as well as extensions into multiobjective optimization. Application of heuristic programming methods using evolutionary and genetic algorithms are described, along with application of neural networks and fuzzy rule-based systems for inferring reservoir system operating rules.

Recently, the United States Bureau of Reclamation agreed to increase spring releases from Upper Colorado River Basin reservoirs to create a more natural flow regime in the lower Gunnison River and Upper Colorado River. Fishery managers have expressed concern that new operations could change reservoir conditions and jeopardize popular and economically important reservoir sport fisheries. This study attempts to predict how one aspect of reservoir conditions, thermal structure, might respond to new dam operations at Blue Mesa Reservoir aimed at addressing downstream ecological concerns. A one-dimensional thermal model (CE-THERM) is applied to simulating thermal effects of “traditional” and proposed “new” dam operation scenarios. To evaluate the relative importance of climate and dam operations a sensitivity analysis of hydrologic (i.e., inflows and starting reservoir elevation) and meteorologic (i.e., air temperature, cloud cover, and dew point temperature) inputs was conducted along with an “extreme” dam operation scenario. Results indicate that reservoir managers at Blue Mesa Reservoir have considerable latitude for new operations without negative thermal consequences. The natural variability of climate and hydrology appear to exert stronger control over reservoir thermal structure than reservoir operations at Blue Mesa.

Evaluation of flood risk in an interior watershed (one protected from flooding by a levee) requires application of watershed, channel, statistical, and economic models in an integrated, coordinated manner. The Strong Ranch Slough and Chicken Ranch Slough watershed in Sacramento, California is such a watershed. We describe herein how we used standard-of-practice models within a coincident-frequency analysis framework to evaluate existing flood damage potential and flood damage potential with a variety of proposed damage-reduction measures.

Peak-flow attenuation is widely used in urban storm-water systems. Standard practice typically involves peak shaving of postdevelopment flows to predevelopment peak-flow levels for given return interval storms. In most cases, the flow controls target low-frequency events (storms equal to or larger than the 2-year event), but not smaller, higher-frequency events. More recent emphasis on removal of pollutants in urban runoff has targeted control of high-frequency events (smaller than the 2-year storm). The question of how the time history of discharges from these facilities compares with the predevelopment flow regime has not been addressed. This study examines the effects of the state of practice for urban flow control. Performance is judged on how well the postdevelopment flow frequency curve matches the predevelopment curve. Continuous simulation, using 50 years of hourly rainfall records, was performed on two climatically different locales. Detention ponds with a variety of orifices sized for peak-flow control using traditional and innovative methods were examined singly and in conjunction with extended detention best management practices.

This study demonstrates the use of high-order Pareto optimization (i.e., optimizing a system for more than two objectives) on a long-term monitoring (LTM) application. The LTM application combines quantile kriging and the nondominated sorted genetic algorithm-II (NSGA-II) to successfully balance four objectives: (1) minimizing sampling costs, (2) maximizing the accuracy of interpolated plume maps, (3) maximizing the relative accuracy of contaminant mass estimates, and (4) minimizing estimation uncertainty. Optimizing the LTM application with respect to these objectives reduced the decision space of the problem from a total of 500 million designs to a set of 1,156 designs identified on the Pareto surface. Visualization of a total of eight designs aided in understanding and balancing the objectives of the application en route to a single compromise solution. This study shows that high-order Pareto optimization holds significant potential as a tool that can be used in the balanced design of water resources systems.

Regulatory policy analyses are often based on the results of computer-intensive models that have limitations resulting from their complexity, size, and run-time requirements. This paper describes and applies a reduced-form model (RFM) of a large-scale water quality model developed for regulatory decision support. The RFM addresses the needs of decision makers who are interested in assessing uncertainty in model estimates and developing prescriptive applications of the model. This paper describes the RFM and demonstrates its applicability to four U.S. states that vary in environmental and socioeconomic characteristics. An application to combined sewer overflow (CSO) policy is developed to illustrate how the RFM can improve decision support. Economic benefits of CSO controls are simulated using the RFM and compared with the U.S. Environmental Protection Agency’s control cost estimates. The sensitivity of these benefits to assumptions of the benefit-cost analysis is tested. In terms of environmental decision making, the RFM reveals that it is more important to resolve what loading rates are most appropriate for benefit-cost analysis than it is to precisely model wet-weather hydrology.

Genetic algorithm (GA) optimization is well suited for optimizing the operation of water distribution systems, especially large and complex systems. GAs have good initial convergence characteristics, but slow down considerably once the region of optimal solution has been identified. In this study the efficiency of GA operational optimization was improved through a hybrid method which combines the GA method with a hillclimber search strategy. Hillclimber strategies complement GAs by being efficient in finding a local optimum. Two hillclimber strategies, the Hooke and Jeeves and Fibonacci methods, were investigated. The hybrid method proved to be superior to the pure GA in finding a good solution quickly, both when applied to a test problem and to a large existing water distribution system.

A study of water consumption data collection and processing in the context of on-line monitoring and burst detection in water distribution networks is presented using a methodology for positioning of flow meters and selection of a monitoring time step. A stochastic model analyzes the propagation of initial uncertainty in water consumption through a water supply network and its influence on calculated flows. The model combines EPANET and a Monte Carlo based random sampling procedure that simulates the usage of domestic appliances. The model is validated on a medium-sized water distribution network, located in the southeast region of England. Statistical analysis of the calculated flows suggests that flow meters should be located at pipes with the highest variance in flow in order to make sudden changes in monitored flow detectable. As a result, an appropriate size of a district metering area (DMA) can be proposed. For the case study network studied here, a compromise DMA size of 250 properties is suggested.