Leak detection and condition assessment for water distribution pipelines using fluid transient waves

Abstract

The focus of this PhD research is to develop non-invasive and cost-effective techniques for assessing the structural condition of pressurised pipelines using fluid transient pressure waves. The specific objectives include the detection of leaks and localised deterioration that is distributed along a pipeline, such as extended sections of corrosion or the spalling of cement lining. The latter is described by pipeline condition assessment in this thesis. The transient behaviour of a leak is studied in the frequency domain. Numerical studies conducted in this research demonstrate that two leak-induced patterns (on the resonant and the anti-resonant responses) can exist in a frequency response diagram (FRD). The amplitudes of the responses are related to the impedance of the valve in a reservoir-pipeline-valve (RPV) system. A new leak detection technique has been developed in this research based on the further understanding of the leak-induced patterns. This technique uses the relative sizes of the first three resonant responses to determine the location and size of a single leak in RPV systems. In reservoir-pipeline-dead end systems, the information required for single event leak detection is further reduced to the first two resonant responses. A new measurement strategy for the extraction of the FRD of single pipelines is proposed in this research. The boundary valve loss is used to adjust the amplitude of the leak-induced pattern on the resonant responses and also the sharpness of the resonant peaks. A specific type of pseudo-random binary sequence (PRBS) termed the inverse repeat sequence (IRS), is used as the excitation signal. The antisymmetric property of IRS enables part of the nonlinear responses of the system under excitation to be cancelled out, yielding a measured FRD close to the theoretical linear system response. A side-discharge valve based transient generator is designed and fabricated in this research to implement the new FRD measurement strategy. Laboratory experiments are conducted on an intact pipeline and a pipeline with a leak. This research also conducts analysis of the characteristics of distributed pipe wall deterioration and develops new detection techniques. In a measured pressure trace, the size of the reflection resulting from a section of pipeline with a change in wall thickness is indicative of the characteristic impedance of this section. Once the impedance of this section is determined, the wave speed and wall thickness can be estimated. A technique for the detection of a single deteriorated section in pipelines is developed based on the above analysis. Two other condition assessment techniques are developed to deal with the complexities induced by multiple deteriorated sections. The first technique is termed reconstructive MOC (method of characteristics) analysis, which uses the pressure trace measured at the upstream face of the valve in a RPV system to determine the distribution of the impedance along the pipeline. The algorithm reconstructs a MOC grid by calculating the MOC compatibility equations backwards in time, estimating the properties of the pipeline (impedance, wave speed) and the length of each pipe reach as discretised by the MOC grid from the valve towards the reservoir. Preliminary experimental verification is conducted to verify the applicability of the new technique. The second technique is reconstructive transient analysis (RTA), which can be conducted at any interior accessible points along a pipeline, and does not require a RPV boundary condition. The RTA uses two pressure transducers in close proximity to measure two transient pressure traces in one test. A signal processing algorithm is developed to extract the directional transient waves (traveling upstream and downstream). The use of the directional transient waves enables the step response function (SRF) of the section of pipe upstream or downstream of the paired pressure transducers to be obtained. The reconstructive MOC analysis is then adapted to interpret the SRF to yield the distribution of the impedance, from which the location and severity of distributed deterioration can be identified.