Combustion systems will remain important in hard-to-abate sectors during the energy transition, particularly in industrial heating, power generation and fuel-flexible energy systems. At the same time, these systems must operate with higher efficiency, lower emissions and greater flexibility toward alternative fuels such as hydrogen, ammonia and hybrid fuel blends. This creates a major engineering challenge: combustion processes are highly coupled, multi-physics and strongly dependent on operating conditions, while the internal state of the system cannot be directly monitored with conventional sensing approaches. Current high-fidelity simulation tools such as CFD are accurate but too computationally expensive for real-time monitoring, optimisation or control, especially for advanced combustion regimes such as MILD combustion.
INVENT was conceived to address this gap through a digital twin framework for combustion systems that combines high-fidelity numerical simulations, sparse experimental measurements, dimensionality-reduction techniques, and machine-learning models. The overall objective was to move this technology closer to industrial application by generating and validating datasets for relevant combustion systems, demonstrating practical digital twin functionality, and extending the original framework with new capabilities that improve prediction accuracy, physical consistency, and dynamic forecasting. The project also aimed to assess how the technology could support industrial needs in design and operation, in particular by reducing simulation time and costs, enabling virtual sensing in inaccessible zones, anticipating failures, improving energy efficiency, and supporting lower-emission operations. In this way, the project contributes to the broader goals of industrial digitalisation and decarbonisation by enabling faster and more informed optimisation of combustion-based systems.
