Clustering and Forecasting Model for Water Distribution Systems Analysis

Abstract Submission: The operation and management of water distribution systems rely on water metering devices installed throughout the network for billing, operational, or maintenance purposes. However, urban development and population growth can hinder water providers from uniformly collecting water demand data across the network. As a result, predictive methods are unavailable for these unmetered areas, making them difficult to predict and operate. A typical strategy to handle missing or malfunctioning metering is to understand consumption patterns by utilizing information from similar, measured areas. However, constant patterns cannot capture the dynamic real-time changes thus limited in accurately representing the network behavior. This study introduces a two-step method to partition a water network into similar demand areas in order to predict demand conditions in unmeasured regions. The clustering step employs network topology and hydraulic characteristics to minimize demand variation across clusters. Next, a predictive model is trained using demand data from multiple measured clusters to forecast demand in an unmeasured cluster. Our two-step method reports notable performance when trained on a set of ten clusters from a benchmark hydraulic network. The developed predictive model for unmetered areas offers water providers an alternative approach to understanding demand dynamics across the entire network.

Learning Objectives/Expected Outcome (Optional) :