Lanterna Rally

Author: Alaa Salman

Publisher:

Published: 2011

Total Pages: 195

ISBN-13:

Canada's civil infrastructure systems have been in use for over 79% of their expected service life. Municipalities in Canada have noted that 59% of their water systems needed repair and the condition of 43% of these systems is unacceptable. Therefore, a significant volume of rehabilitation projects are necessary to improve infrastructure performance. Reliability and criticality assessments (RCA) as well as the ability to determine the most suitable methods of rehabilitation are urgently needed in order to allocate the available budget efficiently. The research presented in this thesis aims at developing a priority index (PI) for intervention that considers the combination of RCA for water networks. Sound techniques are utilized to develop the PI such as reliability theory, simple multi-attribute rating technique (SMART), and Analytical Hierarchy Process (AHP). The reliability assessment encompasses two levels: (1) segment and (2) subnetwork reliabilities. The priority index (PI) for intervention is crucial to schedule segment rehabilitation. Simple Multi Attribute Rating Technique (SMART) is used to select the most suitable methods of rehabilitation for these components. Selection of a rehabilitation method is based on several factors: (1) technical feasibility, (2) whether the selection is contractually acceptable, (3) cost iv effectiveness, (4) environmental impact, and (5) whether the rehabilitation method is a new technology or not. The output of rehabilitation selection model is the method of rehabilitation for components coupled with the associated costs and durations for rehabilitation activities for each sub-network. The final stage of this research is to schedule these rehabilitation activities. Scheduling of the rehabilitation activities related to water main networks depends mainly on available budget and planning time. Other factors, such as network reliability, criticality, location, contract size, and rehabilitation method(s), also affect the scheduling process. This research presents a method for optimizing the scheduling of rehabilitation work for water distribution networks. The method utilizes unsupervised neural networks (UNNs) and Mixed Integer Non Linear Programming (MINLP) and performs the scheduling in two stages. In the first stage, UNNs are used to group water mains according to their locations and rehabilitation methods. In the second stage, MINLP is used to determine the number of rehabilitation contract packages and to generate an optimized schedule based on these packages considering network reliability, criticality, contract size, and planning time. Data on water network are collected from the city of Hamilton, Ontario, Canada. Four sub-networks are selected randomly from the entire network to represent four types of land use; undeveloped, residential, park, and commercial/industrial. The data is used as a test bed to validate and demonstrate the use of the developed research methodology. An automated tool (DSSWATER), based on the developed methodology, is developed to assist users and decision makers. The developed models and tools are expected to be beneficial to municipal engineers and managers as well as to academics.