**
Reconstruction
of Hydraulic Management of a**

**
Water
Distribution System Using Optimization**

**M.
M. Aral, J. Guan, M. L. Maslia, J. B. Sautner**

**
Multimedia
Environmental Simulations Laboratory**

**
School of Civil
and Environmental Engineering**

**
Georgia Institute
of Technology**

**
Atlanta, Georgia
30332**

An epidemiologic study of childhood leukemia and central nervous system cancers that occurred during the period 1979 through 1996 in Dover Township, N.J., is under study. This study is exploring a wide variety of possible risk factors, one being the exposure to groundwater contaminants that occurred through private and community water supplies (i.e., the water-distribution system serving the area). For this purpose, a model of the complex water-distribution system has been developed and calibrated through an extensive field investigation. The components of this water-distribution system, such as number of pipes, number of tanks, and number of supply wells in the network, have changed significantly over a 35-year period (1962 through 1996) - the time frame established for the epidemiologic investigation. For the completion of the epidemiologic study, information on monthly distribution of water in the network, based on a management strategy is necessary for the period 1962 through 1996. For certain months of the study period, some data are available to estimate the operational characteristics and the management strategy employed to operate the water-distribution system. For other months, there are no data to predict the characteristics of this management system. Further complicating the study is that within a given month of any year, the water-distribution system may have operated under peak-, winter-, or average-demand conditions. Manual reconstruction of this management system is time consuming, labor intensive, and costly, given the complexity of the system and the time constraints imposed on the study. In an effort to reduce the required computational time, the problem was formulated as an optimization problem. For each month of the investigation, the management strategy was arrived at by obtaining a solution to the optimization problem. Thus, in this study, it is assumed that the water-distribution system was operated in an optimum manner at all times to satisfy the minimum and maximum pressure constraints and tank level constraints of the system. Given these assumptions, Genetic Algorithms along with the EPANET water-distribution network solver, is used to solve the optimization problem and to develop the historical management strategy used in the water-distribution system, serving the Dover Township area, NJ. This process reduced the required solution time significantly and generated a historically consistent management strategy for the system.