Water Distribution System Burst Detection Using a Nonlinear Kalman Filter

Abstract

A water distribution system burst from a sudden pipe failure results in water loss and disruption of customer service. Artificial neural networks, state estimation, and statistical process control (SPC) have been applied to detect bursts. However, system operational condition changes such as the set of operating pumps and valve closures greatly complicates the detection problem. Thus, to date applications have been limited to networks that are supplied by gravity or under consistent operation conditions. This study seeks to overcome these limitations using a nonlinear Kalman filter (NKF) method to identify system condition, estimate system state, and detect bursts.