Effective Utilisation of Pressure Transients in Pipeline Systems

Abstract

Fluid transients are potentially devastating phenomena in pipelines. However, some of their attributes can bring benefits to various areas of research. This thesis explores three different effective ways to use transients. The considered applications include pipeline condition assessment, unsteady flow measurements and investigation on the dynamic behaviour of hydraulic systems. The analyses involve numerical simulations by a Method of Characteristics and transfer matrix models and experimental investigations using the laboratory pipeline system at the University of Canterbury. An impulse response function (IRF) as a transient-based condition assessment technique is studied experimentally in this thesis, and potential challenges with its use in real systems are identified and methods for improvement are presented. The proposed frequency-domain based technique achieves some enlargement and sharpening of the IRF pulses, but the extent of enhancement is constrained by a low-pass filter for noise removal. Another proposed method yields a cleaner and sharper IRF and the IRF pulses become more detectable because of multi-scale cross correlations and a wavelet-based denoising technique. A flow metering technique known as the Kinetic Pressure Difference method is verified in order to assess its applicability in various flow conditions. The observed errors in the numerical and experimental studies are comparable to the accuracy of the existing commercial flow meters in steady flow measurements, demonstrating the potential of this method. It is identified that the system wave speed is the critical parameter for accurate flow measurements. An additional study is conducted to confirm the ability of the method to capture the change in a steady flow. Transients are also used in this thesis for the investigation on the dynamic characteristic of leaks. Two approaches are taken for the experimental investigation. One method assesses the frequency dependent effects from the dilation of the leak reflection in the time domain. For all leak sizes considered, no significant dilation is observed, implying the absence of frequency dependency. The other approach observes the change in the frequency components of the transient signals after interacting with leaks. A comparison between the results with and without the leak exhibited some difference. Between the two approaches, the one based on the frequency-domain observations makes use of a greater amount of information from the transients and is expected to represent the unsteady leak characteristics more precisely.