Page 104 - IJEEE-2023-Vol19-ISSUE-1
P. 104

Received: 16 November 2022              Revised: 14 January 2023  Accepted: 15 January 2023
DOI: 10.37917/ijeee.19.1.13
                                                                                             Vol. 19| Issue 1| June 2023

                                                                                                         Ð Open Access

Iraqi Journal for Electrical and Electronic Engineering

Original Article

Machine Learning Approach Based on Smart Ball
COMSOL Multiphysics Simulation for Pipe Leak

                          Detection

                                              Marwa H. Abed*, Wasan A. Wali, Musaab Alaziz
                                 Department of Computer Engineering, University of Basrah, Basrah, Iraq

Correspondence
*Marwa H. Abed

Department of Computer Engineering,
University of Basrah, Basrah, Iraq

Email: engpg.marwa.abd@uobasrah.edu.iq

Abstract
Due to the changing flow conditions during the pipeline's operation, several locations of erosion, damage, and failure occur.
Leak prevention and early leak detection techniques are the best pipeline risk mitigation measures. To reduce detection time,
pipeline models that can simulate these breaches are essential. In this study, numerical modeling using COMSOL Multiphysics
is suggested for different fluid types, velocities, pressure distributions, and temperature distributions. The system consists of 12
meters of 8-inch pipe. A movable ball with a diameter of 5 inches is placed within. The findings show that dead zones happen
more often in oil than in gas. Pipe insulation is facilitated by the gas phase's thermal inefficiency (thermal conductivity). The
fluid mixing is improved by 2.5 m/s when the temperature is the lowest. More than water and gas, oil viscosity and dead zones
lower maximum pressure. Pressure decreases with maximum velocity and vice versa. The acquired oil data set is utilized to
calibrate the Support Vector Machine and Decision Tree techniques using MATLAB R2021a, ensuring the precision of the
measurement. The classification result reveals that the Support Vector Machine (SVM) and Decision Tree (DT) models have
the best average accuracy, which is 98.8%, and 99.87 %, respectively.
KEYWORDS: Leakage Detection, Fluid Flow, Heat Transfer, Pipeline Monitoring, SVM, DT, Computational Fluid
Dynamics.

                        I. INTRODUCTION                           acoustic sensors, satellite surveillance, mass and volume
                                                                  balance, and analytical model-based procedures, have been
   According to the Worldwide World Energy Report in              put into practice. These methods depend on many aspects of
2020, natural gas and fossil fuels accounted for 30% of the       the process, such as the temperature, pressure, mass and
global demand for energy production[1]. Fuel delivery,            volumetric flow rates, and so on [8].
including fuel supply through fluid pipes, is essential to
energy provision [2]. Applied stress, environmental factors,         According to Dong et al. [9], the most beneficial of these
and noise levels are among the many challenges that               technologies is the negative pressure strategy since it offers
pipelines encounter, with mechanical systems deteriorating        great leak sensitivity and availability. Unfortunately, this
to various extents [3]. Certain factors in the leakage            method has a high likelihood of creating a false alert if the
phenomenon are challenging to measure in actuality. Some          pressure measurement records indicate significant changes
researchers have recently embraced computational fluid            or if the leak is tiny (0.5% of nominal flow) [10]. As a result,
dynamics as a technique to aid in this process. Many              it needs to investigate using numerical methods into the
industrial operations involve the simultaneous flow of two        hydrodynamics of heavy oil-water flow in a vertical pipe
immiscible liquids in vertical pipes, one of which is the         with a slight leak, which is much more challenging to detect
petroleum industry [4]. Because of the importance of the          using conventional methods [11]. Pipelines that are utilized
topic, many authors have focused their attention on the study     in real-time operations are frequently situated in extreme
of methods that could be used to identify leaks in pipelines      environments, such as those found in the sea, where they
that are used for the production and transit of oil [5]–[7]. At   have been subjected to the pressure that is exerted by the
the current time, many different techniques for detecting         water; in the middle of the desert; or even underground,
leaks, including those based on harmful pressure waves,           where they are subjected to the force that is exerted by the

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and

reproduction in any medium, provided the original work is properly cited.

© 2023 The Authors. Published by Iraqi Journal for Electrical and Electronic Engineering by College of Engineering, University of Basrah.

https://doi.org/10.37917/ijeee.19.1.13                                                                   https://www.ijeee.edu.iq 100
   99   100   101   102   103   104   105   106   107   108   109