Pipeline leak detection is a matter of great interest for companies who transport petroleum and its derivatives, in face of rising exigencies of environmental policies inindustrialized and industrializing countries. However, existing technologies are not yet fully consolidated and many studies have been accomplished in order to achieve betterlevels of sensitivity and reliability for pipeline leak detection in a wide range of flowing conditions. In this sense, this study presents the results obtained from frequency spectrumanalysis of pressure signals from pipelines in several flowing conditions like normal flowing, leakages, pump switching, etc. The results show that is possible to distinguish between the frequency spectra of those different flowing conditions, allowing recognition and announce of liquid pipeline leakages from pressure monitoring. Based upon these results, a pipeline leak detection algorithm employing frequency analysis of pressure signals is proposed, along with a methodology for its tuning and calibration. The proposed algorithm and its tuning methodology are evaluated with data obtained from real leakages accomplished in pipelines transferring crude oil and water, in order to evaluate its sensitivity, reliability and applicability to different flowing conditions
展开▼
