Leak detection and location of flanged pipes: An integrated approach of principle component analysis and guided wave mode

Leak detection and location of flanged pipes: An integrated approach of principle component analysis and guided wave mode

Leak detection

Variational mode decomposition

Principal component analysis

Petroleum transportation

Learning systems

Acoustic emission testing

Guided electromagnetic wave propagation

As flanged pipes have a crucial role in oil and gas transportation, accidental leaks and their detection are of great importance, and represent a great concern in safety management of pipes. The present study develops an integrated approach, based on acoustic emission (AE) signals, in order to detect and locate leaks in flanged pipes. It is depicted into two main steps: leak detection (leak regional location) and precise leak location. The proposed method adopts the principal component analysis (PCA) as a semi-supervised learning method, relying on the Hotelling's T2 statistic and the Q statistic (squared prediction error (SPE)) as the residuals, to detect the leak occurrence. Once the leak pipe segment is located, the precise leak location adopts variational mode decomposition (VMD) to decompose the leak AE signals to obtain the domain frequency band for determining the wave speed in combination with the dispersive curve of guided wave. Furthermore, de-noised signals can be obtained by reconstructing the intrinsic mode functions (IMF) of the VMD based on the energy ratio and after the basic cross-correlation is calculated to obtain the time delay of the de-noised signals. For illustrative purposes, an experimental setup, which includes 5 different leak apertures (10, 12, 15, 20 and 27 mm), is used for investigation. The experimental validations show that the PCA-based method can detect and locate a leak in a certain pipe segment. In addition, the precise leak location errors are all in acceptable interval values ranging from 0.8% up to 12.1%. 2020 Elsevier Ltd

129

Diao, Xu

Chi, Zhaozhao

Jiang, Juncheng

Mebarki, Ahmed

Ni, Lei

Wang, Zhirong

Hao, Yongmei

Safety Science

2020

journalArticle

9257535

10.1016/j.ssci.2020.104809

http://dx.doi.org/10.1016/j.ssci.2020.104809