Journal of Water Resources Planning and Management

A nonlinear decision‐control model for real‐time operation of a dual‐purpose water system for electrical energy generation and irrigation is presented. Maximal return depends on the implemented flow‐rate forecasting method and varying costs for energy sold to an external utility. The results of two forecasting models and the historical data are compared, indicating the importance of an accurate data base for supporting the decision towards optimal operation of time‐dependent water systems.

Network flow programming (NFP) is used to simulate the operation of the Oroville reservoir system of the California State Water Project (SWP). The simulation uses three integrated NFP models developed and calibrated to simulate the hydraulic operation, hydropower generation, and contractual obligations of the California Department of Water Resources (CDWR) to deliver energy to Southern California Edison (SCE): (1) The Oroville Hydraulic Network (ORONET) is designed to meet SWP demands downstream of the Oroville reservoir system and target storage in the complex; (2) the Oroville Periodic Hydraulic Network (OROPRD) determines the hydraulic feasibility of a desired electrical generation pattern at the Oroville complex given the boundary conditions provided by ORONET; and (3) the Thermalito Afterbay Network (THANET) simulates the periodic operation of the Oroville complex to determine energy entitlement SCE. The simulation is performed on weekly and daily levels, with up‐to‐hourly detail. This paper demonstrates a real‐world application of NFP to water resources and an application of a traditional optimization technique to screen and simulate reservoir operations.

A strategy for prioritizing decisions for the maintenance of a water distribution system is developed. Using component and network reliability based on time‐varying connectivity concepts, the probabilities that water will be available at demand points in the system are calculated to determine a reliability surface. At any time, this surface is used to locate low reliability areas, which identify parts of the system that need maintenance priority. The specific components that must be repaired or replaced are determined using a component importance criterion that measures the overall effect of component maintenance on the system reliability. The procedure is applied to one hypothetical and two real‐life water distributions systems.

This paper describes a mixed integer programming model developed for the water management of Southern California Edison's Big Creek hydroelectric facility. It is an application paper—a review of past and present practice of an electric utility company engaged in the management of California's water resources. The model is deterministic with monthly decision variables and annual optimization horizons. The objective is the maximization of the total value of the produced energy. The model is primarily intended for long‐term planning purposes. However, the addition of a new power plant in 1987 reduced the guidance value of the recorded pattern of historical operations that have been improved based on dispatchers' experience over a long period of time. Consequently, the model is found to be very useful in determining short‐term (annual or semiannual) operating policies. Its principal output is a water‐release policy during on‐ and off‐peak periods for 12 consecutive months.

The storm‐water management model (SWMM) is applied to a 10‐sq‐mi urbanized residential area. The application is intentionally constrained to a rather gross spatial scale (no modeling of storm sewer transport) in order to mimic anticipated regional modeling efforts. Three different levels of watershed spatial abstraction are investigated in concert with the impact of using two different parameters for model calibration (pervious depression storage and percent imperviousness) along with three separate calibration events. The calibrated SWMM model performed quite well at predicting both total runoff volume and peak flow rate. Successful application was most sensitive to calibration event selection, with the smallest event resulting in the worst overall performance. The importance of watershed conceptualization level was less significant, with no noticeable difference between the two highest abstraction levels. Use of percent imperviousness, as the single model calibration parameter, was generally more successful than adjustment of pervious depression storage. SWMM water‐quality simulation was also quite acceptable, but with a consistent underestimation of in‐stream total suspended solids concentrations.

The recurrence interval is the expected length of time intervening between consecutive geophysical events. The interest in this work is in episodic (catastrophic or cataclysmic) phenomena such as large floods or rare earthquakes, i.e., events that have a long‐term recurrence meaningful in the scale of geologic time. Using probabilistic reasoning, the relationship between the recurrence interval and the distribution of geophysical events, the concept of risk, the memoryless property of the distribution of the time to the next episode, and the behavior of the “excess” time to the next event are examined. Also examined is the contrast between the necessary statistical conditions required for the mathematical analysis of the aforementioned variables, and geoclimatic changes inherent to natural processes that give the classical analysis of the recurrence of extreme geophysical phenomena and the risk presented by their potential occurrence a somewhat limited value. Our analysis includes annual and partial duration series.

Selection of rainfall‐runoff models is dependent on the modeling objectives and the simulation accuracy required. Output reliability is proposed as a measure to help in choosing the appropriate model for a given rainfall‐runoff model use. Output reliability can be expressed by the output variable cumulative distribution function, probability density function, confidence limits, standard deviation, or probability of acceptable percent deviation from the predicted value. Approximations of these reliability measures could be established by methods such as first‐order second‐moment reliability analysis. To demonstrate the approach, an example is given of estimating peak discharge for a watershed in central Illinois using both the HEC‐1 and RORB rainfall‐runoff models.