Balaji Mohan
Lead Scientist
Saudi Aramco
Dr. Balaji Mohan is a Lead Scientist at Saudi Aramco's R&D Center. He holds a PhD from the National University of Singapore. His expertise includes optimizing internal combustion engines and alternative fuels. Currently, he applies Artificial Intelligence to reduce the carbon footprint of internal combustion engines and develop customized fuels for advanced engine systems. Dr. Mohan has authored over 45 journal publications and 30 conference contributions, and has filed 4 patents on innovative AI-based solutions and fuel designs.
Participates in
TECHNICAL PROGRAMME | Energy Fuels and Molecules
Alternative Fuels - E fuels, Biofuels and SAF
Forum 15 | Digital Poster Plaza 3
29
April
11:30
13:30
UTC+3
The increasing urgency to mitigate climate change has intensified the search for sustainable fuels, particularly in the aviation sector, where decarbonization remains a formidable challenge. Sustainable aviation fuels (SAFs), particularly e-fuels, present a promising avenue for reducing greenhouse gas emissions associated with air travel. However, the intrinsic variability in the chemical composition and properties of e-fuels poses substantial barriers to their widespread adoption. To overcome these hurdles, it is crucial to identify suitable additives that can augment critical fuel properties such as energy density, combustion efficiency, and thermal stability.
In response to this challenge, our research leverages the power of machine learning (ML) to predict and optimize the properties of SAFs. Utilizing the Chemical SuperLearner (ChemSL) framework, we demonstrate the effectiveness of ML models in estimating key aviation fuel properties, including density, net heat of combustion, cetane number, boiling point, and freezing point. The ChemSL framework integrates multiple molecular representations with a SuperLearner ensemble model, allowing for the construction of highly accurate and robust property prediction models.
Our results show that the ChemSL models achieve high predictive accuracy, as evidenced by low mean absolute errors and high R-squared values for all considered properties. Furthermore, the diversity of base learners included in the ChemSL models underscores the complexity of the prediction tasks and highlights the importance of employing ensemble methods to capture intricate relationships within the data.
This study paves the way for the rapid identification of suitable additives to enhance the performance of sustainable aviation e-fuels, thereby contributing to a more efficient and environmentally friendly transition towards sustainable aviation. Future research directions include expanding the molecular database to screen additives compatible with middle distillates and exploring the applicability of the ChemSL framework to other challenging problems in fuel design and optimization.
In response to this challenge, our research leverages the power of machine learning (ML) to predict and optimize the properties of SAFs. Utilizing the Chemical SuperLearner (ChemSL) framework, we demonstrate the effectiveness of ML models in estimating key aviation fuel properties, including density, net heat of combustion, cetane number, boiling point, and freezing point. The ChemSL framework integrates multiple molecular representations with a SuperLearner ensemble model, allowing for the construction of highly accurate and robust property prediction models.
Our results show that the ChemSL models achieve high predictive accuracy, as evidenced by low mean absolute errors and high R-squared values for all considered properties. Furthermore, the diversity of base learners included in the ChemSL models underscores the complexity of the prediction tasks and highlights the importance of employing ensemble methods to capture intricate relationships within the data.
This study paves the way for the rapid identification of suitable additives to enhance the performance of sustainable aviation e-fuels, thereby contributing to a more efficient and environmentally friendly transition towards sustainable aviation. Future research directions include expanding the molecular database to screen additives compatible with middle distillates and exploring the applicability of the ChemSL framework to other challenging problems in fuel design and optimization.
