Towards the autonomous molecular design and characterisation of lubricant base oils and additives
File(s)
Author(s)
Ogbomo, Mark
Type
Thesis or dissertation
Abstract
Improving the performance of lubricant formulations is critical for enhancing the energy efficiency and durability of machine components.
As internal combustion engines are progressively replaced by electric motors, lubricant formulations must meet new operational demands, such as improved thermal stability, electrical compatibility, and reduced environmental impact.
Despite these needs, the discovery and optimisation of lubricant formulations remains largely empirical and dependent on costly trial-and-error experimentation.
This thesis attempts to address these challenges through the development and application of computational methods to accelerate the molecular-level characterisation and design of lubricant base oils, and provide a deeper understanding of the nanoscale behaviour of a prevalent lubricant additive.
The first part of this thesis focusses on using molecular dynamics (MD) simulations to explore the mechanochemical decomposition mechanisms of tricresyl phosphate (TCP), a widely used anti-wear additive, under varying surface chemistries, temperatures, sliding speeds, and shear stresses.
The second part of this thesis explores the use of machine learning (ML) models for the prediction of hydrocarbon transport properties, assessing the impact of molecular representation, model selection, and dataset generation using MD simulations.
The performance of these ML models is benchmarked against experimental measurements and MD simulations, and their limitations in generalising to out-of-distribution data are examined.
Finally, a multi-objective optimisation framework is developed that integrates interpretable ML with evolutionary algorithms to propose novel base oil molecules with enhanced properties compared to current industrial lubricant base oils.
The methods and findings presented in this thesis contribute to a deeper molecular-level understanding of lubricant base oil and additive behaviour, and demonstrate the potential of computational approaches to guide the design of next-generation high-performance and environmentally-friendly lubricants.
As internal combustion engines are progressively replaced by electric motors, lubricant formulations must meet new operational demands, such as improved thermal stability, electrical compatibility, and reduced environmental impact.
Despite these needs, the discovery and optimisation of lubricant formulations remains largely empirical and dependent on costly trial-and-error experimentation.
This thesis attempts to address these challenges through the development and application of computational methods to accelerate the molecular-level characterisation and design of lubricant base oils, and provide a deeper understanding of the nanoscale behaviour of a prevalent lubricant additive.
The first part of this thesis focusses on using molecular dynamics (MD) simulations to explore the mechanochemical decomposition mechanisms of tricresyl phosphate (TCP), a widely used anti-wear additive, under varying surface chemistries, temperatures, sliding speeds, and shear stresses.
The second part of this thesis explores the use of machine learning (ML) models for the prediction of hydrocarbon transport properties, assessing the impact of molecular representation, model selection, and dataset generation using MD simulations.
The performance of these ML models is benchmarked against experimental measurements and MD simulations, and their limitations in generalising to out-of-distribution data are examined.
Finally, a multi-objective optimisation framework is developed that integrates interpretable ML with evolutionary algorithms to propose novel base oil molecules with enhanced properties compared to current industrial lubricant base oils.
The methods and findings presented in this thesis contribute to a deeper molecular-level understanding of lubricant base oil and additive behaviour, and demonstrate the potential of computational approaches to guide the design of next-generation high-performance and environmentally-friendly lubricants.
Version
Open Access
Date Issued
2025-06-23
Date Awarded
2025-10-01
Copyright Statement
Attribution-NonCommercial-ShareAlike 4.0 International Licence (CC BY NC-SA)
Advisor
Ewen, James
Dini, Daniele
Sponsor
Engineering and Physical Sciences Research Council
Grant Number
EP/T51780X/1
Publisher Department
Department of Mechanical Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)