Research On Leak Detection For Pipeline

Pipeline transportation is convenient, high efficient and environment-friendly. Together with railway, highway, airway and shipping it becomes one of the five important transportations. Pipeline leakage is a frequent phenomenon due to inevitable aging, erosion and contrived damage, which causes serious economic loss and environmental pollution as well. Therefore the real-time leakage detection of pipeline is an important research topic around the world. In this thesis, a transient flow model for pipeline leakage is put forward based on transient flow theory. The transient flow model is based on finite volume method and the numerical scheme is Godunov scheme. Computing program for transient flow model is developed in Fortran language. The time series of pressure can be obtained for any specified station along the pipeline. Pressures for different leakage flowrates and leakage stations are obtained in various scenarios. The analysis of pressure under different leakage condition shows that the peak pressure in time series depends on both leakage location and flowrates. The closer the leakage station and the measured station are, the greater the peak pressure value is. For the same leakage station, the peak pressure value increase with the decrease of leakage flowrate. Fast Fourier transform is used to convert the time series of pressure into frequency domain for the study of leakage frequency feature. The frequency response diagram under different leakage conditions are analysis. The influence of different leakage conditions on the peak value in frequent response diagram is summarized. To obtain the detailed information of leakage-the leak location is assumed, the time series and frequent response diagram are obtained, the sum of squire of difference between the theoretic value and the "measured value"is calculated and plotted with the assumed leak location as sensitivity analysis curve. The location of minimum in the curve indicates the leakage location. Comparison shows that the time series is more accurate in sensitivity analysis than the frequency response diagram in locating leakage. In both cases, the accuracy increase with the increase of leakage flowrate. When the leakage flowrate is small, the pressure curve in frequency domain loses its sensitivity first then that in time domain loses its sensitivity. Analysis shows that the locating error is within 1% of the pipeline length when the flowrate of leakage is 1% of the steady flowrate. The limitation of present model is discussed and the suggestion for further study is put forward according to the limitation of model and the conclusion obtained from previous calculation.