Journal of Water Resources Planning and Management

Pumping costs are the major operating expense of ground‐water supply systems. This paper presents a nonlinear optimization model to minimize pumping costs for both a well field and a main water‐supply distribution system. Considerations are given to individual well losses, pump efficiencies, and the hydraulic losses in the pipe network. In addition, the transient drawdown at each well is included in a well‐field model. When the demand served is less than the total capacity, there is a potential for reducing costs in the selection of pumps to meet that demand. A simulation model in conjunction with an optimization algorithm is assembled and optimized using the general nonlinear optimization program MINOS. For a given demand, the optimization procedure provides the best combination of pumps to meet that demand. Two example problems are given to evaluate the validity of the underlying assumptions and to demonstrate some of the characteristics of the proposed procedure.

A water‐quality‐data‐management survey of 200 water‐quality agencies in the United States was conducted in 1991. The survey solicited information from each agency regarding the methods of data storage and management used, the type of activities involving water‐quality data that were performed by the agency, the types of water‐quality data used, the sources of the data used, and interagency activities involving water‐quality data. The survey results indicated that there is a high demand for data sharing, since almost all agencies responded that their water‐quality data are used by others. There is also a potential for data sharing because of the predominant use of ARC/INFO geographic information systems and dBASE database software. Future software development for water‐quality data applications should use this potential and be appropriate for IBM‐compatible personal computers.

Demand‐management policy rules are sought during drought and impending drought for a water system consisting of a reservoir dedicated only to water supply. The creation of such rules requires solution of a nonlinear, nonseparable mathematical programming problem. A polytope search algorithm using a combination of simulation and optimization is compared to an iterative mixed integer programming method to determine the parameters of continuous demand management rules. The signal used for calling rationing is a trigger volume given in terms of months of demand (as a volume) that are needed in storage. When the sum of actual storage plus anticipated inflow is less than the trigger volume, rationing is initiated. The extent of rationing or demand reduction that is required is determined by the ratio of the sum of storage plus inflow to the trigger volume. The two methodologies for parameter determination are compared using as a criteria the maximum shortage that occurs over some planning period.

A mathematical model is presented for the purpose of representing the washoff of sediments from small urban watersheds during storm events. The model simulates the contribution of sediments from each of the different land surfaces present within an urban watershed. It has been determined that different land surfaces contribute sediments differently, and that a better estimate of the total sediment load could be determined by representing each of the major land surfaces independently. The model was developed and tested using water‐quality data from four small urban watersheds in Houston and Austin, Tex., and by using data from bare‐soil erosion plot studies. The performance of the model was determined by comparing the simulated results with the observed results during the storm events that were analyzed. Given the appropriate parameters, the model provides reasonable estimates of the observed sediment yield from each of the different land surfaces studied. The rate at which sediments are washed off the impervious area of a watershed is shown to correlate with the length of time since rainfall has last occurred.

A multisite river‐flow forecasting model is presented in this work for the Han River Basin, Korea. Seventeen locations on the Han River are simultaneously considered for real‐time flow forecasting. A linear stochastic model based on the ARMAX class of models is developed for each forecasting location, and the Kalman filter is used to forecast and update optimal estimates of the flows. Two other filters are used to update the model parameters and noise statistics in real time. The structure of the models, in flexible black‐box forms, provide for several exogenous inputs, including precipitation, antecedent soil moisture effect, natural inflows from upstream subcatchments, and controlled releases from reservoirs. This work extends the real‐time forecasting models introduced by Awwad and Valdés in 1992 to include meteorological terms essential to carrying out multiple‐step‐ahead hydrologic forecasting. It introduces the rainfall‐runoff process as part of an adaptive stochastic model, with the catchment's response coefficients updated on‐line as information becomes available. The software developed for this application is part of and compatible with a comprehensive model being developed by the U.S. Army Corps of Engineers and the Han River Flood Control Center for the Han River Basin.

This paper describes a new methodology that can select the pipes to be rehabilitated and/or replaced in an existing water‐distribution system and determine the increase in pumping capacities so that the water demand and pressure requirements at all demand nodes are satisfied while the total rehabilitation and energy cost is minimized. Four cost functions are considered: pipe replacement cost, pipe rehabilitation cost, pipe repair cost, and pumping cost. The methodology considers the trade‐offs among decisions regarding each pipe: replace pipes, reline pipes, or leave as is. The mathematical formulation is a mixed‐integer nonlinear programming problem. The solution procedure combines an implicit enumeration scheme using a branch and bound algorithm along with a generalized reduced gradient procedure to solve a nonlinear subproblem. The nonlinear subproblem is solved by interfacing a hydraulic simulator (KYPIPE) with the GRG2 optimizer in which the pipe diameter and pump horsepower are the decision (control) variables.

The Texas Estuarine Mathematical Programming (TxEMP) model was developed as a tool to study the effect of freshwater inflow to the bays and estuaries in Texas and to help establish longterm objectives of maintaining and enhancing an ecologically healthy environment for the estuaries. The model is a nonlinear, stochastic, multiobjective mathematical programming model. It represents needs of the biological system through the use of salinity‐inflow regression equations and fishery‐harvest‐inflow regression equations. If quantitative information is available, nutrient and sediment requirements can be taken into account. Hydrological information is incorporated as the monthly lower bounds and upper bounds on inflows. Salinity constraints and fishery harvest constraints are treated as chance constraints. The objective function can be inflow minimization or maximization, fishery‐harvest maximization or minimization, maximization of satisfying the salinity constraint, or satisfying the fishery‐harvest constraint. A multiobjective approach is taken to resolving the estuarine management problem by presenting results as a performance curve that contains a series of solutions ranging from a minimum required inflow, maximum harvest solutions, and a maximum allowable inflow. This paper presents mathematical details of the TxEMP model and an actual application to determine the release policy from the upstream reservoirs to the Nueces Estuary.

Modeling catchment runoff requires spatial discretization of the catchment for computation. This study investigates the effect that the size of the computational discretization has on the results of a simulation. Three different types of models (namely, a rational formula model, an Australian nonlinear conceptual model (RORB), and a physical‐process‐based distributed model (HEC‐1)) were tested by applying them to a hypothetical catchment. For each model the spatial discretization size for computations was systematically decreased, while all other catchment and model parameters were held constant. It was observed that even for a simple, homogeneous, hypothetical catchment the simulated runoff results are significantly affected by the computational spatial size of the catchment used regardless of the type of model applied to the catchment.