Comparison between leak detection methods for open water channels

Abstract

In this paper we present three methods for detecting the presence of a leak in an open water channel. The available measurements are the water level and the gate position at the upstream and downstream end of a channel reach. The first method uses the mismatch between the data and a model of the channel as an indicator of the presence of leaks. The second method uses an extended Kalman filter to estimate the size of the leak while the third method makes use of a Luenberger type observer to estimate the size of the leak. All three methods make use of the cumulative sum (CUSUM) algorithm to detect the presence of a leak. The methods have been tested on real data from the Coleambally Channel no. 6 in Australia, and a comparison between the methods is presented in terms of leak detection performance and computational cost.