Multi-objective Stochastic Design of Water Distribution Systems
Author | : Yves R. Filion |
Publisher | : |
Total Pages | : 472 |
Release | : 2006 |
ISBN-10 | : 0494157550 |
ISBN-13 | : 9780494157558 |
Rating | : 4/5 (50 Downloads) |
Book excerpt: This thesis presents a new framework named the Multi-objective Stochastic Design Method (MOSDM) that can be used to guide the design and rehabilitation of water distribution networks. In the design process, the MOSDM seeks to answer three fundamental questions. What loads can come to bear on a water distribution network? How frequently will a water distribution network fail when subjected to these loads, and how quickly on average will it recover from failure? And when loads cause a network to fail, what are the consequences and damages? To answer these three questions, a Monte Carlo Simulation is used to simulate hydraulic failures in a network subject to stochastic loads (e.g., demand, fire flows, and pipe breaks) and to update probabilistic measures of reliability, resilience, expected annual damages, and the probability of catastrophic failure. The last two measures are developed in this thesis to characterize the human and economic consequences and damages of hydraulic failure in distribution networks. A case study shows that the MOSDM can identify design alternatives with a low damage level and a low probability of catastrophic failure. The MOSDM is also used to develop an intensity-duration-frequency methodology to guide the rehabilitation of cul-de-sacs---a major component in most networks---in a manner that balances the duration (measure of severity) and frequency of hydraulic and water quality failures. In anticipation that the MOSDM may be applied to design real networks in the future, an uncertainty analysis is undertaken to investigate how, and to what extent, uncertainties in the covariances of demand---the most important load category in most networks---affect network costs (capital and operating) as well as the network performance assessed with the probabilistic measures of reliability, resilience, and expected annual damages. Preliminary results suggest that whether uncertainties in the covariances of demand exert a significant influence on cost and performance in design depends in large part on the presence of tank storage.