Burst Detection and Location in Pipelines and Pipe Networks - With Application in Water Distribution

Abstract: Sudden pipe bursts occur in high-pressure water transmission pipelines and water distribution networks. The consequences of these bursts can be very expensive due to the outage time while the burst pipe is repaired, the cost of repair, and damage to surrounding property and infrastructure. As a result, it is advantageous to minimise the detection and location time after the burst occurs. Currently, there is no effective solution for the burst detection and location problem in water distribution systems. The applications in oil and gas pipelines and pipe networks show the advantages of continuous monitoring. A number of techniques are used to determine the location and size of a burst. One of the most promising approaches is fluid transient modelling and analysis. Pressure transient analysis has also been successfully applied for detection and location of existing leaks. A sudden pipe burst creates a negative pressure wave that travels in both directions away from the burst point. The analysis of this wave is the main principle for the techniques presented in this thesis. Experiences from previous research suggest that, to achieve the best performance, single pipelines and pipe networks have to be treated separately. Thus, two different approaches for burst detection and location are presented. When a burst occurs in a pipeline, the burst-induced pressure wave travels in both directions along the pipeline and is reflected at the boundaries. Using a pressure trace measured at one location along the pipeline, the timing of the initial and reflected burst-induced waves determines the location of the burst. The presented continuous monitoring technique uses the modified two-sided cumulative sum (CUSUM) algorithm to detect abrupt changes in the pressure data caused by the pipe break. The results from both laboratory and field pipelines are used to verify the proposed method. Different burst and measurement locations are tested. The results are promising for burst detection and location in real systems. In the network case, continuous pressure measurements at two locations are analysed. The burst detection and location algorithm is based on the difference between the arrival times of the burst-induced pressure wave at each measurement point and on the measured wave magnitude. The arrival times are determined automatically in real time. A method for determining optimal measurement locations is also presented. Results from numerical simulations show that the proposed technique has potential as a tool for effective detection and location of bursts in real pipe networks. Most transient-based techniques use transient modelling for analysis or validation. The development of a transient model comprises part of the work presented in this thesis. The Method of Characteristics is used to solve the governing unsteady flow equations. Models for unsteady friction, leakage and burst are included. The model is used to validate the results of burst detection and location. iv