Multi-modal identification of leakage-induced acoustic vibration in gas-filled pipelines by selection of coherent frequency band

Highlights

• The acoustic coupling between in-pipe gas and pipe wall is so weak that the leak signal is mainly shell dominated wave.

• The leakage-induced shell dominated wave has multi-modal and dispersive characteristics which results in a larger leak location errors.

• The single mode can be extracted by selection of coherent frequency band, which will provide an effective way to reduce leak loaction errors.

Abstract

Leak location using correlation-based acoustic method is a common practice in gas-filled pipelines, which will be extremely effective when the leakage-induced acoustic wave has a constant propagation speed. This premise is often not satisfied as result of multi-modal and dispersive nature which will cause acoustic speed to vary with frequency and modal type. In this work, the multi-modal identification based on selection of coherent frequency band is proposed for improving leak location in gas pipelines. The dispersive and multi-modal characteristics are studied by theory analysis and experimental validation to give understanding of the reason that correlation-based leak location has a larger error at times. The results demonstrate that the multi-modal type can be identified from law of acoustic speed variation by selection of coherent frequency band and multi-modal and dispersive characteristics are the main cause of a larger leak location error.

Introduction

The pipelines are the most important way of transporting fluid from one place to another, such as oil, gas, water, etc. As a result of corrosion, aging or third-party damage, pipe leakage occurs from time to time [1], which leads to not only the loss of natural resources but also serious environmental pollution and threat to public safety. Therefore, it is essential to detect and locate a leakage using a real-time online method to prevent and mitigate the consequences.

Many studies have been conducted on leak detection [[2], [3], [4], [5], [6], [7], [8]], ranging from regular manual inspections to automated measurements using various types of sensors. Among them, acoustic-based leak detection technology is recognized as an effective method to implement a true real-time leak monitoring system owing to its high leak-detection speed and ease of retrofitting [9]. It is reported that the dynamic pressure sensors are used to pick up the gas leakage-induced acoustic waves for leak detection and location in the gas pipelines [10]. But the dynamic pressure sensors need to be inserted into the gas pipelines for leak detection, which is quite inconvenient in practical application. Sun et al. demonstrate that the acoustic emission (AE) sensors mounted on the outer surface of the gas pipelines can be used to conveniently acquire the gas leakage-induced acoustic waves [11]. These previous studies indicate that direct picking up the gas leakage-induced acoustic waves on the outer surface of the pipelines utilizing AE sensors is a more convenient and more feasible for leak detection and location in the gas pipelines.

In the leak location surveys using AE technique, the most useful method for locating a leakage is based on the cross-correlation of the two measured acoustic signals, on either sides of a leak [[12], [13], [14], [15]]. The correlation-based leak location is effective on condition that the noise interference of two measured leakage signals is not correlated and the signal to noise ratio (SNR) is high [16]. In practice, the measured data contains correlated noise and the SNR is low when the leakage-induced acoustic wave travels along the gas pipeline, which results in a larger time-delay estimation (TDE) errors [[17], [18], [19], [20]]. In order to suppress noise interference, the generalized correlation-based leak location method is proposed in water supply pipelines, which relies on prior knowledge of signals and noise [21]. Leak detection using AE technology belongs to passive detection method which cannot satisfy the requirements leading to a larger TDE errors based on generalized correlation. Hence, adaptive TDE-based leak location is proposed in water pipelines which can obtain a correct time delay without any prior knowledge of signals and noise [16].

So far, many efforts have been focused on TDE algorithm to reduce leak location errors under low SNR of measured data. However, leak location error is still large using the accurate time delay from time to time. The TDE-based leak location is effective which requires that the acoustic speed has been known and constant. In practice, the requirement is not always satisfied because the leakage-induced acoustic waves are multi-modally and dispersive blended signals. Furthermore, distinct modes possess different dispersions, which will generate varying acoustic speed with frequency and modal type, consequently resulting in missing detection and location of a leakage. In this work, the modal and dispersive characteristics are investigated by theoretical and experimental analysis to understand the reasons that TDE-based leak location has a larger error at times. The characteristic frequency band of measured data in gas pipelines is determined using coherence analysis. Then, the measurement method of acoustic wave speed using phase difference of cross spectrum is proposed in this study. The acoustic wave speed can be abstained by the time delay of leak signal in characteristic frequency band to explore speed change law in gas pipelines, which can make clear that the TDE-based leak location has a larger error at times due to multi-modal and dispersive characteristics.