Multi-component HyChem kinetic mechanism generation using trust-region Bayesian optimization

The current regulations aimed at reducing the aviation carbon footprint promote the extensive use of sustainable aviation fuels (SAFs). Yet, how the peculiar physico-chemical properties of these synthetic jet fuels affect combustion phenomena in aeronautical burners remains an open question. In this regard, to foster an understanding of the impact of novel SAFs on combustor operability figures of merit, we propose an extension of the Hybrid Chemistry (HyChem) approach for efficiently generating kinetic mechanisms for the oxidation of multi-component fuel surrogates via trust-region Bayesian optimization (TuRBO). This methodology enables the numerical investigation of alternative jet fuel behavior in practical combustion systems by adopting multi-component surrogates and accounting for the role of preferential evaporation on flame characterization. The fuel pyrolysis steps are modeled with a set of seven lumped reactions for each component, which are merged with a detailed mechanism for the oxidation of the pyrolysis products. We test this methodology on a two-component physico-chemical surrogate for the alcohol-to-jet (ATJ) Jet C-1, underscoring the trade-off between computational efficiency and design space exploration offered by the TuRBO algorithm, which employs local refinements of the design space to tune the parameters describing the pyrolysis reactions, compared with other available optimization techniques. The results exhibit a good alignment with the experimental results and the original HyChem single-component approach. The importance of a multi-component surrogate description is confirmed by examining how the fuel composition evolves along the distillation curve and how this affects ignition. These analyses of preferential evaporation and its impact on flame stabilization and combustion regimes support the broader adoption of SAFs in commercial aviation. • A new extension of HyChem for multi-component fuels is proposed. • The approach allows preferential evaporation to be accounted for in combustion analyses. • Trust-region Bayesian optimization enables a rapid exploration of the design space. • A kinetic mechanism for a two-component surrogate of C-1 is developed.