Robust Water System Design with Commercial Intelligent Search Optimizers

Abstract

Intelligent-search-based optimizers have shown promise in providing improved links between analysis and design. Genetic-algorithm-based optimizers are often used, but other heuristic methods, such as tabu search, have also been used with good results. A major impediment to using heuristic search methods has been the lack of user-friendly, commercially available software. This is no longer the case due to the availability of several commercial intelligent-search-based optimizers. Robust analysis and optimization of a water distribution network is demonstrated with four commercial optimizers. The hydraulic analysis is performed with WinPipes.EXE, a Windows program based on the EPANET source code. The method is demonstrated by optimizing the New York City Water Supply Tunnel problem, and by the optimal design of a 15-loop, Almos water distribution system. The method is robust because it uses reliable and efficient commercial optimizers, a popular pipe network solver, and constraint penalties to meet the multiple goals of a reliable, low-cost water distribution system, capable of meeting maximum hour demands and fire flow demands.