Smart WDS management: Pipe burst detection using real-time monitoring data

Abstract

Recently, advanced and smart techniques are being implemented for improving water distribution system (WDS) management and control. Those methods are mostly based on field data measured in real-time throughout the system of bigdata characteristics especially with respect to its volume and velocity. An interesting research issue is to investigate how to extract useful information from big data for efficient WDS management and control (e.g., pipe burst and leakage detection). This study applies the Western Electric Company (WEC) method, a statistical process control method, for pipe burst detection which plots field data measured in real-time around control limits obtained from historical normal field measurements. We investigate the impact of meter location and the number of meters on pipe burst detectability (i.e., detection probability and false alarm rate). Control and out-of-control pipe flow data are synthetically generated by using a hydraulic model of the Austin network and simulating pipe bursts under stochastic demand conditions.