Enhancing Performance and Reducing Emissions in Natural Gas Aspirated Engines through Machine Learning Algorithm

Moinuddin Ansari, Mohammed Ameer

Enhancing Performance and Reducing Emissions in Natural Gas Aspirated Engines through Machine Learning Algorithm

dc.contributor.advisor	Kazempoor, Pejman
dc.contributor.author	Moinuddin Ansari, Mohammed Ameer
dc.contributor.committeeMember	Merchan Merchan, Wilson E
dc.contributor.committeeMember	Ghamarian, Iman
dc.date.accessioned	2024-01-03T17:28:42Z
dc.date.available	2024-01-03T17:28:42Z
dc.date.issued	2023-12-15
dc.date.manuscript	2023-12
dc.description.abstract	In an era where the global energy landscape is increasingly defined by the dual imperatives of efficiency and sustainability, the natural gas sector stands at a crucial juncture. The engines powering this sector, especially Natural Gas Fired Reciprocating Engines (NGFRE), are well known for their performance as well as considerable emissions, posing a stark challenge to environmental sustainability goals. This thesis addresses this pivotal issue, presenting a machine learning-based solution to optimize NGFRE performance while substantially reducing their environmental footprint. The research is anchored in an experimental framework involving the AJAX DPC-81 engine compressor, evaluated across a spectrum of operational loads from 40% to 75%. The study leverages an extensive array of sensors to collect detailed real-time data on engine performance, emissions, and vibration parameters. Central to the methodology is the strategic adjustment of the Air Management System (AMS), varying air/fuel ratio to explore their impact on engine dynamics and emissions. The study also incorporates a comprehensive vibration analysis, providing critical insights into the engine's operational stability under different load conditions. Machine Learning (ML) techniques, including Linear Regression, Artificial Neural Networks (ANN), and Support Vector Machines (SVM), are integrated with a Programmable Logic Controller (PLC). This integration not only facilitates a nuanced analysis of the collected data but also enables the accurate prediction of engine performance, paving the way for real-time adaptive control systems. The findings of this research are both revealing and impactful. A notable instance is observed at a 40% engine load with a 70% bypass valve opening, where emissions of methane (CH4) plummet by 64%, nitrogen oxides (NOx) by 52%, and Volatile Organic Compounds (VOC) by 50%. This substantial decrease highlights the effectiveness of the ML-driven approach in curbing harmful emissions. Further, the study unveils the manipulation of the bypass valve position can lead to enhanced fuel efficiency and improved engine stability. For example, at a 75% engine load, the research demonstrates that optimal emission reduction is achieved with a mere 10% bypass valve opening, illuminating the delicate interplay between engine load parameters and environmental emissions. In conclusion, the study demonstrates the effectiveness of ML in enhancing NGFRE performance. It sets a foundation for developing intelligent engine systems that can self-adjust for optimal performance and minimal environmental impact, forging a path to a future where the two are seamlessly integrated.	en_US
dc.identifier.uri	https://hdl.handle.net/11244/340092
dc.language	en_US	en_US
dc.rights	Attribution-NonCommercial-NoDerivatives 4.0 International	*
dc.rights.uri	http://creativecommons.org/licenses/by-nc-nd/4.0/	*
dc.subject	Machine learning	en_US
dc.subject	Emissions	en_US
dc.subject	Reciprocating engines	en_US
dc.subject	Oil and gas	en_US
dc.thesis.degree	Master of Science	en_US
dc.title	Enhancing Performance and Reducing Emissions in Natural Gas Aspirated Engines through Machine Learning Algorithm	en_US
ou.group	Gallogly College of Engineering::School of Aerospace and Mechanical Engineering	en_US
shareok.orcid	https://orcid.org/0000-0002-8377-3820	en_US