A METHOD FOR CONSTRUCTING DMAs BY A GRAPH CLUSTERING METHOD AND GENETIC ALGORITHMS

Abstract

 This paper proposes a method for creating District Metered Areas considering minimization of the life cycle cost (total of a capital cost and an operational cost). To minimize the cost, some mathematical approaches are applied. Our approach considers a distribution pipe network as a graph network model. Under the model, a graph clustering algorithm decides boundaries of DMAs. Then, genetic algorithm decides entering pipes of DMAs and pump heads. We applied our proposed method to a virtual pipe network to construct DMAs. Compared with the case considering only a capital cost, the result obtained by our method reduces the life cycle cost by 9%. Moreover, we tried sensitivity analysis by fluctuating water demands in the network. By this trial, we confirmed that DMAs robust to hydraulic condition can be created by setting high lower bound on water pressure in a network.