Leak location detection in gas pipeline network

• Main Author: Mahato Lipika
• Other Authors: Justin Dauwels
• Format: Theses and Dissertations
• Language: English
• Published: 2017
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/69525
• Institution: Nanyang Technological University

Pipeline networks are the cheapest and safest mode of transportation of oil and natural gas all over the world. In Singapore, natural gas is used to generate electricity hence, real time monitoring of the gas pipeline network is crucial for uninterrupted electricity production. The distribution pipeline network of Singapore is underground, low pressure system which is a challenge and unexplored field of research. Currently deployed leak detection methods in commercial sectors require a lot of resources, manpower and sometimes obstruction of the natural operation of the pipelines for manual monitoring. Gas leakage from pipelines can cause not only huge financial losses but also major accidents. However, the complexity of these networks due to large underground infrastructure makes manual detection of leakage almost impossible. This project contributes to the problem by implementing and verifying the pressure profiles generated by presence of leak in the network. Leak is said to have occurred when a sudden peak of pressure is seen in the pressure difference profile. The pressure profiles also depict how leaks near to the source of inlet are easy to locate than the leaks that occur far from the source. The aim of the thesis is to locate the leak even in the presence of noise and analyse the different methods for leak detection. A physical approximate model for stationary study is considered and then the data generated by COMSOL is worked on in MATLAB. The simulation graphs show the results of mean absolute error against Gaussian noise for single inlet value. The study is extended further, by considering multiple inlet values. By multi-dimensional scaling, pipe with leak can be predicted with high accuracy. These obtained results can be utilised in future research for an extended network and developing predictive leak localisation models for the research project.