الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 262 |  |  |
| | | from | 262 | | |
Abstract
A computer stochastic model program (GACCnet) was developed for the optimal network design constrained by reliability for water distribution system analysis which was formulated as an optimization problem under uncertainty. The optimal design problem was formulated as a single-objective chance constraint minimization problem restricted with a pre-specified level of uncertainty and solved using the genetic algorithm (GAs) technique with the Newton-Raphson method used for hydraulic simulation. The sources of uncertainty analyzed here were the future nodal external demands, the pipe roughness, and the nodal minimum pressure head requirements which were assumed to be random normally distributed variables with given means and standard deviations. The node and system reliabilities were then evaluated for the least-cost design solution of the network using the Monte Carlo simulation technique. The model was then designed to be efficiently used for both design of new networks and rehabilitation of existing one as well. The results of the proposed model were illustrated by the application of designing a two-loop pipe network and El-Mostakbal City pipe network. The model was also applied to the rehabilitation of the New York City water supply tunnels problem. All the networks upon which the model was applied did not contain any pumps or valves. The obtained results for two different values of the coefficient of variation of 10% and 20% showed that the design of water distribution networks using stochastic approaches rather than deterministic one caused a significant decrease in cost and a better performance under future uncertain operating conditions. Also, the chance constraint formulation for the optimization of water distribution systems is more effective in reducing the computer time especially when applied to relatively large systems. The system and node reliabilities were found to be proportional to the total cost of the network for a fixed value of the coefficient of variation which means that the higher the reliability, the higher the cost of the network. High reliability of the system means better performance of the network under normal conditions. Finally, the final perfect accepted decision in this regard depends on the judgment of the well trained greatly qualified design engineer.