Journal of Water Resources Planning and Management

An epidemiologic study of childhood leukemia and central nervous system cancers that occurred in the period 1979 through 1996 in Dover Township, N.J., is being conducted. Because groundwater contamination has been documented historically in public- and private-supply wells, there is the possibility of exposure through this pathway. The Dover Township area has been primarily served by a public water supply that relies solely on groundwater; therefore, a protocol has been developed for using a water-distribution model such as EPANET as a tool to assist the exposure assessment component of epidemiologic investigation. The model is being used to investigate the question of human exposure to groundwater contaminants. Because of the unavailability of historical data, the model was calibrated to the present-day (1998) water-distribution system characteristics. Pressure data were gathered simultaneously at 25 hydrants throughout the distribution system using continuous recording pressure data loggers during 48 h tests in March and August 1998. Data for storage tank water levels, system demand, and pump and well status (on/off) were also obtained. Field data gathering procedures, calibration results, and water-quality simulation using a naturally occurring element (barium), as well as an analysis indicating the percent of water originating from points of entry to the water-distribution system for 1998 conditions, are presented.

Dynamic (extended period simulation) models have largely replaced steady-state models for simulation of water quality in distribution networks. Dynamic modeling generally has been limited to periods of a day or a few days. This paper describes three modeling studies that used long time-series analyses to characterize contamination in the water distribution system of a major city. Continuous runs of the EPANET model, each covering a period of one year, were used to develop matrices that could be used to calculate average annual dosages at each node in the distribution system for a wide variety of contaminant concentrations at the source. The matrix approach provided a way to facilitate integration of ground-water modeling results with distribution modeling. The use of long-term runs overcomes the shortcomings of steady-state modeling or use of “representative days” in predicting long-term average concentrations.

A new methodology has been developed for determining the optimal operation of water distribution system pumps with water quality considerations. The methodology is based upon describing the operation as a discrete time optimal scheduling problem that can be used to determine the optimal operation schedules of the pumps in distribution systems. The solution methodology is based upon a mathematical programming approach resulting in a large-scale nonlinear programming problem that cannot be solved using existing nonlinear codes. The solution of the optimization problem is obtained by interfacing a hydraulic and water quality simulation code, EPANET, with a nonlinear optimization code, GRG2. Bound constraints on the state variables are incorporated into the objective function using the augmented Lagrangian penalty method. Three objective functions were used in the model to minimize (1) the deviations of actual substance concentrations from desired concentration values; (2) the total pump duration times; or (3) the total energy cost. The effectiveness of the methodology was tested using a hypothetical water distribution system. All three objective functions were capable of finding an optimal pumping schedule with water quality considerations.

The very marked chlorine decay measured in one section of the distribution system of a large North American city and the fluctuations in this decay were studied. The results obtained from operational data collection, flow measurements, and EPANET hydraulic modeling demonstrate that low chlorine residuals in this particular distribution system section cannot be attributed to any exceptional consumption by corrosion particles, loose deposits, or biofilms. On the contrary, the hydraulic configuration of this distribution system area (tank design and pumping cycles) results in such high residence times in one tank (5.6–7.9 days) that the low chlorine residuals observed are self-explanatory. Detailed analysis of the results indicates that the principles of hydraulic (dys)function in this zone are well described by the model, which can be relied upon to compute and compare several design or operational solutions to minimize residence times. Two of these solutions were field tested and were found to be equally efficient in diminishing the age of water. As a consequence, detectable chlorine residuals were restored in the area.

In a multisource, multiquality regional water distribution system, water agencies often find that it is necessary to impose blending requirements at certain control points in the system in order to secure the desired water quality downstream of the control points. A nonlinear multicommodity flow model is proposed to optimize water delivery in such a system while at the same time meeting the imposed blending requirements. Waters from different sources with different water quality are considered as different commodities concurrently sharing a common distribution system. The model incorporates a set of nonlinear constraints to account for the perfect mixing condition at nodes where waters from multiple sources merge. The proposed model was first tested and verified on a simplified hypothetical system and then applied to a large-scale regional water distribution system in Southern California. The model is steady state and monthly. The results obtained indicate that the proposed methodology can efficiently control water blend in a complex water distribution system while minimizing the total shortage in water delivery.

Transient conditions in closed conduits have traditionally been modeled as 1D flows with the implicit assumption that velocity profile and friction losses can be accurately predicted using equivalent 1D velocities. Although more complex fluid models have been suggested, there has been little direct experimental basis for selecting one model over another. This paper briefly reviews the significance of the 1D assumption and the historical approaches proposed for improving the numerical modeling of transient events. To address the critical need for better data, an experimental apparatus is described, and preliminary measurements of velocity profiles during two transient events caused by valve operation are presented. The velocity profiles recorded during these transient events clearly show regions of flow recirculation, flow reversal, and an increased intensity of fluid turbulence. The experimental pressures are compared to a water hammer model using a conventional quasi-steady representation of head loss and one with an improved unsteady loss model, with the unsteady model demonstrating a superior ability to track the decay in pressure peak after the first cycle. However, a number of details of the experimental pressure response are still not accurately reproduced by the unsteady friction model.

Today, most municipal water utilities use computerized numerical models of their distribution networks. These models must be calibrated to the physical system if they are to provide accurate results. Calibration entails adjusting certain model parameters, usually the pipe roughness coefficients, until the model accurately predicts the results of a series of flow tests. The choice of flow test locations, the sampling design, is currently done on an ad hoc basis. If the flow test locations are less than optimal, the data collected may yield insufficient information for an accurate calibration, leaving the modeler a choice between collecting more data, at additional expense, or using a less accurate model. This paper describes the use of a genetic algorithm to optimize the sampling design. The genetic algorithm was applied to a network model for a small town in Ohio and shown to perform extremely well, matching the optimal solutions produced by complete enumeration in a series of validation tests.

In this paper a parallel computing based software demonstrator for the simulation and leakage minimization of water networks is presented. This demonstrator, based on the EPANET package, tackles three different types of problems making use of parallel computing. First, the solution of the hydraulic problem is treated by means of the gradient method. The key point in the parallelization of the method is the solution of the underlying linear systems, which is carried out by means of a multifrontal Choleski method. Second, the water quality simulation problem is approached by using the discrete volume element method. The application of parallel computing is based on dividing the water network in several parts using the multilevel recursive bisection graph partitioning algorithm. Finally, the problem of leakage minimization using pressure reducing valves is approached. This results in the formulation of an optimization problem for each time step, which is solved by means of sequential quadratic programming. Because these subproblems are independent of each other, they can be solved in parallel.

Leak detection and calibration of pipe internal roughnesses in a water distribution network are significant issues for water authorities around the world. Computer simulation of water distribution systems to determine the location and size of leaks is emerging as an important tool. A major uncertainty in developing computer models is the condition of the interior of the pipes in the network, especially if they are old. An innovative technique for leak detection and calibration called the inverse transient technique has been recently developed. This paper uses the genetic algorithm (GA) technique in conjunction with the inverse transient method to detect leaks and friction factors in water distribution systems. A continuous variable representation has been developed for the GA coding scheme in this paper. Two new GA operators for crossover and mutation are also introduced. The inverse transient method using the GA technique is effective at finding leakage locations and magnitudes while simultaneously finding the friction factors for different transient data record lengths.