Diagnosis of a Leaky Pipeline Carrying Multiphase Flow under Plug Flow Conditions

Abstract

Multiphase flows are crucial to the oil and gas industry since most petroleum companies produce and transport both gas and oil simultaneously. Pipeline leaks are frequently caused by corrosion, aging, and metal deterioration. After an incident, the energy sector not only loses money but also raises environmental and safety concerns. Therefore, developing a successful tool for instantaneous leakage identification in pipelines becomes crucial. In the current work, a leaky pipeline carrying multiphase flow is numerically simulated using Ansys-Fluent under plug flow conditions. The obtained numerical results were validated against experimental data collected from an experimental setup. After that, Probability Density Function (PDF), Wavelet Transform (WT), and Empirical Mode Decomposition (EMD) methods were applied to the obtained time series signals. On the other hand, the analysis is complemented by the application of several machine learning models like Random Forest (RF), Support Vector Machine (SVM), and k-Nearest Neighbors (k-NN). For instance, it is observed that the Empirical Mode Decomposition exhibits better performance in leakage identification