Towards digital twinning of turbulent reacting flows: from microscale to extreme events

The drive towards sustainable aircraft propulsion is requiring drastic changes in propulsion systems, with the use of carbon-free fuels such as hydrogen or ammonia. These fuels have significantly different reacting properties than traditional jet fuel, which make their safe and stable control within the combustor turbulent environment very challenging as both fuels can trigger extreme unsafe changes in the combustor state. Hydrogen can trigger flashback, where the flame suddenly moves back in the injector potentially destroying it, whereas ammonia can result in blowoff, where the flame suddenly extinguishes, unsafely shutting down the engine. In addition, both fuels are susceptible to thermoacoustic instabilities, where high amplitude pressure waves are generated, damaging the turbine blades. All these types of events are called extreme events as they constitute unpredictable sudden large changes in the system state away from its normal state. To tackle this complex challenge, the synergistic use of high fidelity simulations with recent data-driven modelling and digital twinning techniques have shown a great potential in providing the physical insights into and predictive models of such extreme events, paving the way for their prevention. In this seminar, I will illustrate how such an approach can be used to obtain such predictive models at different scales: from reaction/turbulence interaction at the microscale, to flashback precursors identification for online monitoring and thermoacoustic instabilities prediction at the combustor design stage.