Physics-informed reduced-order modeling for digital twins of sustainable combustion systems – PROMCOM

In the current energy transition scenario, energy-intensive industries and long-range transportation sectors remain heavily reliant on the combustion of fossil fuels. Thus, renewable and alternative fuels, such as hydrogen, ammonia, and biofuels, are vital for achieving carbon neutrality in these difficult-to-decarbonize sectors. Advanced combustion technologies are needed to convert these fuels efficiently while minimizing harmful emissions. Current experimental and numerical methods, including Computational Fluid Dynamics (CFD), are either too limited in scope or computationally expensive for design and optimization procedures targeting advanced industrial-scale combustion systems. To address this gap, the project proposes the development of Reduced-Order Models (ROMs), which can simplify complex combustion phenomena into manageable input-output relationships and enable the development of digital twins of combustion systems. The project aims to develop non-intrusive projection-based ROMs via physics-informed machine learning. Data-driven, physics-based approaches will be applied to experimental and numerical data of a lean premixed prevaporized (LPP) gas turbine model combustor. The project is organized around three work packages (WPs): WP1 concerns the use of feature extraction techniques to find the best low-dimensional representation of the combustion system; WP2 involves the development of an adaptive simulation framework relying on clustering approaches to enable feasible and accurate numerical simulations; WP3 deals with generating simulation-based prognostic ROMs featuring regression models able to capture the system dynamics in the low-dimensional feature space. The proposed final deliverable will be a digital twin that can predict the flame responses observed in the gas turbine model combustor and the transition between them, validating a methodology that will enable the real-time monitoring, control, and optimization of sustainable combustion systems.