|Ph.D Student||Sela Polina|
|Subject||Water Distribution Systems Aggregation|
|Department||Department of Civil and Environmental Engineering||Supervisor||Professor Avi Ostfeld|
|Full Thesis text|
A water distribution system (WDS) is a collection of hydraulic control elements (e.g. pipes, pumps, tanks, valves, etc.) designed to deliver prescribed quantities of water from sources (e.g. reservoirs, wells) to consumers at desired pressures and qualities. WDS analysis of the fully detailed network can result in thousand to tens of thousands of hydraulic components making the solution of management and control problems extremely complex. This research is aimed at developing methodologies for aggregation and simplification of water distribution systems to reduce the complexity of water distribution systems and management problems being analyzed while retaining the properties of the original full-scale system. The methodologies are applied to real large-scale networks concentrating on the characteristics and control of hydraulics, water quality, and security of water distribution systems. The four problems addressed in this dissertation are:
1) Aggregation - an all-pipe network model can make model calibration, verification, and system operation simulations time consuming and complex. A method of hydraulic and water quality aggregation to reduce the complexity of the networks is developed and applied. The model outcome provides a reduced network with fewer elements (i.e., nodes and links), closely resembling both hydraulic (i.e., pressures) and water quality (i.e., concentrations) behavior of the original full-sized system.
2) Clustering - to gain insight on the system behavior a methodology based on graph theory is developed and demonstrated by simplifying its operation through topological/connectivity analysis. The algorithm divides the system into clusters according to the flow directions in pipes. The resulted clustering is generic and can be utilized for different purposes such as water security enhancements by sensor placements at clusters or contaminant isolation.
3) Rare events - to protect public health and minimize the community affected by contaminant intrusion, contamination warning systems are being designed to analyze water quality and detect pollutant incursion. The suggested algorithm is able to sample efficiently a rare subset of hazardous contamination events having small probability to occur but an extreme impact on the distribution system.
4) Bayesian network - Bayesian belief networks are a probabilistic analysis tool for representing and analyzing problems involving uncertainty. The presented methodology proposes estimating the likelihoods of the injection location of a contaminant and its propagation in the system using Bayesian network statistics. The data collected from the monitoring stations located at any of the full system nodes is exploited to inquire information about the possible sources of contamination and the consequent polluted nodes.