Optimal Operation of Water Distribution Systems Using a Graph Theory–Based Configuration of District Metered Areas

Abstract

Optimal operation of a large water distribution system (WDS) has always been a tedious task, especially when combined with determination of district metered areas (DMAs). This paper presents a novel framework based on graph theory and optimization models to design DMA configuration and identify optimal operation of large WDSs for both dry and rainy seasons. The methodology comprises three main phases: preliminary analysis, DMA configuration, and optimal operation. The preliminary analysis assists in identifying key features and potential bottlenecks in the WDS, narrowing down the large number of decision variables. The second phase employs a graph theory approach to specify DMAs and adjust their configuration based on similarity of total water demands and pressure uniformity in DMAs. The third phase uses several consecutive, single-objective and multiobjective optimization models. The decision variables are pipe rehabilitation, tank upgrade, location of valves and pipe closures, and valve settings for each DMA. The objective functions are to minimize total annual cost of rehabilitation, water age, and pressure uniformity. The proposed methodology is demonstrated through its application to the large, real-world WDS of E-Town. The results show that the proposed methodology can determine a desirable DMA configuration mainly supplied directly by trunk mains.