Research
Computational methods for energy and propulsion, grounded in physics and informed by data.
We work across four connected themes, combining numerical simulation, scientific machine learning and experimental validation.
Reactive flows and combustion modelling
We develop high-fidelity and reduced-order models of chemically reactive flows, focusing on the interaction between turbulence, chemistry and radiation.
Our work includes flame-speed measurement corrections, quasi-dimensional combustion models for fuel-blend optimisation, and non-equilibrium plasmas for combustion enhancement.
Selected work: Faghih et al., Combustion and Flame (2023); Paykani et al., Fuel (2021).
Scientific machine learning for chemical kinetics
Detailed chemical kinetics is costly to integrate within flow simulations. We develop surrogate models that reduce this cost while retaining physical constraints and estimating prediction uncertainty.
Our methods include Gaussian-process source-term integration and uncertainty-aware deep ensembles that adapt during simulation. We assess accuracy, stability and computational cost against established numerical solvers.
Selected work: Üstün and Paykani, Proceedings of the Combustion Institute (2025); International Journal of Hydrogen Energy (2024).
Hydrogen and ammonia as fuels
We study the combustion of hydrogen, ammonia, methane and syngas blends, using chemical-kinetic analysis and data-driven models to predict burning velocities across compositions and operating conditions.
Current research also investigates liquid-ammonia direct-injection sprays with Kyushu University, and the exhaust and contrail implications of cracked ammonia for aviation.
Selected work: Mashruk et al., Energy & Fuels (2024); Üstün et al., Fuel (2024); Paykani et al., Progress in Energy and Combustion Science (2022).
Thermal management of electric machines
Heat removal is an important constraint on the power density of electric traction motors. We investigate direct oil-jet cooling of hairpin end-windings through CFD and experiments.
We examine nozzle geometry, flow rate and impingement angle, including multi-nozzle configurations, to guide cooling-system design with industrial partners.
Selected work: Maddumage et al., IEEE Transactions on Transportation Electrification (2025); Applied Thermal Engineering (2025).
