|M.Sc Student||Zeidan Mohamad|
|Subject||Multi-Objective Optimization for Trading off Operational|
Cost, Leakage, and Water Age in Water Distribution
|Department||Department of Civil and Environmental Engineering||Supervisor||Professor Avi Ostfeld|
|Full Thesis text|
This research proposes a multi-objective optimization model for minimizing water age, leakage, pressure, and operational costs, under consumer demands and system constraints. The developed methodology links a multi-objective genetic algorithm with EPANET.
The network is first divided into segregated district metered areas (DMAs) allowing to effectively control leakage problems. In addition, segregation of a looped WDS to a number of independent DMAs improves the water distribution network management as it simplifies water balance calculations and reduces water security risks, as contaminants movement in the network are better controlled.
The proposed methodology incorporates the following stages: (1) Mapping the water distribution system: the distribution system is mapped into an undirected graph, G= (V, E) in which the vertices V represent the consumers, sources and tanks, and the edges E represent the connecting pipes, pumps and valves. (2) Communities structure identification: subdividing the graph G into clusters using the community structure method proposed by Clauset et al. (2004) (the method uses modularity as an indicator to the quality of the resulted graph divisions into clusters). (3) Dendrogram cutting: the dendrogram (i.e., the outcome of stage 2) indicates the hierarchical community structure of the water distribution network. It is up to the decision maker to define the limits of the community nodes number or consumers demand, and thus the location of cutting the dendrogram to meet those constraints. (4) Communities isolation: isolating the clusters to reduce the interaction between them and thus limit contaminants movement in the system through closing some of the feed lines that connect between the clusters. The isolation process starts from the smallest bridge pipe in diameter until the last one, closing part of them using a heuristic method. (5) Implanting new pump operation scheduling, using a genetic algorithm to competitively reduce the water leakage, the water age and the operational costs while satisfying the customer requirements for minimum pressure. A multi-objective optimization problem is solved, resulting in a multi-objective Pareto front. The proposed methodology is demonstrated through a couple of case studies of increasing complexity.