A Wavelet Approach to Pipeline Leak Detection by Pressure Analysis

Abstract

One of the most challenging tasks in an oil field is implementation of a software-based leak detection system on a multi-phase flow pipeline. When a leak occurs in a multi-phase flow pipeline, the flow cannot be measured with accuracy. So, none of the various pipeline leak detection methodologies can offer good performance on a multi-phase flow pipeline. This paper will discuss implementation of a leak detection system in a particular oil field using state-of-the-art signal processing techniques to apply to the data collected in a oil pipeline. This leak detection system is still in development and uses a more practical approach to the problem than traditional methods and was implemented on a PC under the Windows operating system. Windowing, joint time-frequency analysis and wavelets were considered to develop methods of detecting leaks by watching for the wavefront. The idea behind these techniques is to cut the signal of interest into several parts and then analyze the parts separately. It is impossible to know the exact frequency and the exact time of occurrence of the leak frequency in a signal. In other words, a leak signal can simply not be represented as a point in the time-frequency space. It is very important how one cuts the signal to implement the analysis. The wavelet transform or wavelet analysis is probably the most recent solution to overcome the shortcomings of the Fourier transform. The wavelet transforms are used to perform atomic decompositions of the pressure signal that comes from a single point of a pipeline. A number of time-frequency decompositions are attempted. What is expected of this decomposition is that it fits the perceptible changes in the pressure and then an Artificial intelligent System (AIS) decides if the variations in the signal are inherent (common-cause variations) or external to the process (failed instrument, occurrence of a leak, causes that are not part of the process). The AIS learns about continual changes in the pipeline. This is useful as pipeline operation always changes and instrument drift could occur over a long time period.