Session: 05-03 Hydrogen-fueled gas turbines

Paper Number: 129279

129279 - A Comparative Analysis of Various Machine Learning Approaches for Fault Diagnostics of Hydrogen Fueled Gas Turbines 

Ever increasing global warming and greenhouse gas emissions motivated the world leaders to devise a framework for mitigation of climate change through the Paris Climate Accord. In line with global regulations and policies, gas turbines that own significant share in power generation in energy sector, need considerable measures for decarbonization.  In this context, gas turbine industry is taking big leaps in meeting the energy transition targets by incorporating low carbon fuels for electric power generation. Hydrogen is among the potential alternative fuel that can bring down carbon emissions significantly.  The utilization of hydrogen in gas turbines has some underlying challenges such as corrosion mainly originating from increased steam content in the hot gas path. In addition to corrosion, the gas turbine compressor is vulnerable to fouling which is most commonly occurring fault in gas turbine operating over certain time window. Both faults are susceptible to performance and health degradation and subsequently trigger downtimes. To avoid expensive asset loss caused by unexpected downtimes and shutdowns, timely maintenance decision making is required. Therefore, simple, accurate and computationally efficient fault detection and diagnostics models becomes crucial for timely assessment of health status of the gas turbines.

The present study is intended to develop a physics-based performance model of a 100-kW micro gas turbine running on 100% hydrogen fuel. The model will be validated with baseline experimental data acquired from test campaigns at University of Stavanger. The data form the experiments and performance model will be utilized for further training the machine learning algorithms. Subsequently, the study aims to identify and test suitable machine learning techniques for developing fault diagnostics models. The techniques might include support vector machine (SVM), Decision Tree, Random Forest algorithm, K-Nearest Neighbors and Artificial Neural Network. The purpose of the study is to explore different methodologies that could help the gas turbine industry in achieving accurate and computationally efficient fault diagnostics of their gas turbine asset in context of hydrogen fuel.

Presenting Author: Muhammad Baqir Hashmi Department of Energy and Petroleum Engineering, University of Stavanger

Presenting Author Biography: Muhammad Baqir Hashmi is currently working as a PhD research fellow at Department of Energy and Petroleum Engineering, in University of Stavanger (UiS) Norway. His current research work involves numerical and experimental performance assessment, diagnostics and prognostics of hydrogen fueled gas turbines.

He has completed his Masters in Mechanical Engineering from Universiti Teknologi PETRONAS, Malaysia. He also worked as Graduate Research Assistant in Mechanical Engineering Department in the same university for two years. His research title was "Transient modeling and intelligent fault diagnostics of variable geometry industrial gas turbines". In UTP Malaysia he worked on projects funded by the local oil and gas company named PETRONAS.

He earned his bachelor degree in Mechanical Engineering from University of Engineering and Technology Lahore, Pakistan. His research work is published in peer reviewed international journals. He owns the computational competencies in MATLAB/Simulink, Python, ANSYS Workbench, GasTurb 12, GSP 11. He routinely reviews the research articles from renowned journals.

His future research interests involve turbomachinery; gas turbines performance; condition monitoring; diagnostics ; prognostics; hydrogen energy; and renewable energy systems

Authors:

Muhammad Baqir Hashmi Department of Energy and Petroleum Engineering, University of Stavanger
Amare Desalegn Fentaye School of Business, Society and Engineering, Mälardalen University
Mohammad Mansouri Department of Energy and Petroleum Engineering, University of Stavanger
Konstantinos G. Kyprianidis School of Business, Society and Engineering, Mälardalen University,