This study focuses on data-driven approaches for burst detection and classifies them into three categories: classification method, prediction-classification method and statistical method. The performance of these methods is discussed. By analysing uncertainty in burst detection, this paper revealed that non-stationary monitoring data and limitations present in these methods challenge the reliability of detection results. Data pre-processing and probabilistic solutions to deal with the uncertainty are summarised. From these findings and discussions, this paper concludes and recommends that: a) data-driven approaches are promising in real-life burst detection and reducing false alarms is an important issue; b) more comprehensive performance evaluation might be necessary, in particular regarding detectable burst size; c) further research on new methods employing multivariate analysis and a new category based on clustering analysis would be beneficial to tackle uncertainty; d) more focus on the use of...

A review of data-driven approaches for burst detection in water distribution systems

Deep learning identifies accurate burst locations in water distribution networks.

Burst detection in district metering areas using a data driven clustering algorithm.

Water loss reduction is important in sustainable water resource management. As one of the main water loss control methods, early detection of hydraulic accidents in district metering areas ...

Burst Detection in District Metering Areas Using Deep Learning Method

Leak detection and location in water distribution systems (WDSs) is of utmost importance for reducing water loss, which is, however, a major challenge for water utility companies. To this end, researchers have proposed a multitude of methods to detect such leaks in WDSs. Model-based and data-driven approaches, in particular, have found widespread uses in this area. In this paper, we reviewed both these approaches and classified the techniques used by them according to their leak detection methods. It is seen that model-based approaches require highly calibrated hydraulic models, and their accuracies are sensitive to modeling and measurement uncertainties. On the contrary, data-driven approaches do not require an in-depth understanding of the WDS. However, they tend to result in high false positive rates. Furthermore, neither of these approaches can handle anomalous variations caused by unexpected water demands.

Review of model-based and data-driven approaches for leak detection and location in water distribution systems

Pipe bursts in a drinking water distribution system lead to water losses, interruption of supply, and damage to streets and houses due to the uncontrolled water flow. To minimize the negative consequences of pipe bursts, an early detection is necessary. This paper describes a heuristic burst detection method, which continuously compares measured and expected values of water demands and pressures. The expected values of the water demand are generated by an adaptive water demand forecasting model, and the expected values of the pressures are generated by a dynamic pressure drop - demand relation estimator. The method was tested off-line on a historic dataset of 5 years of water flow and pressure data in three supply areas (with 650, 11,180 and 130,920 connections) in the western part of the Netherlands. In the period 274 bursts were reported of which, based on the definition we propose in this paper, 38 were considered as relatively larger bursts. The method was able to detect 50, 25.9 and 7.8% in the considered areas related to all bursts, and around 80% in all three areas related to the subset of relatively larger bursts. The method generated false alarms on 3% of the evaluated days on average.

/pdf/heuristic-burst-detection-method-using-flow-and-pressure-5b4yyjloi3.pdf

M. Van De Roer

Papers

Heuristic burst detection method using flow and pressure measurements