A novel chemical clustering approach for the acceleration of 3D CFD simulations of internal combustion engines with complex fuels

Accurate fuel chemistry modelling is critical for internal combustion engine (ICE) simulations, especially when assessing fuel composition effects, preferential evaporation, and emissions. This need grows with the increasing use of e-fuels and blends with conventional fuels. While detailed chemical kinetics offers high fidelity, computational cost often limits its practical application. This paper introduces an extended chemical kinetics clustering approach to accelerate complex fuel simulations in 3D CFD. Building on a previously published method, the approach targets multi-component fuels with strong local inhomogeneities. By grouping cells with similar thermochemical states during source term evaluation, redundant calculations are reduced. The method is robust for a wide range of fuels, from simple surrogates to complex gasoline blends. Benchmark results and an evaluation of the clustering strategy are presented for a direct injection gasoline engine. Simulations are performed using the well stirred reactor approach in Vectis. With a baseline toluene reference fuel (TRF) using the SIP 2.0 chemical mechanism containing 108 chemical species, a tenfold speed up in source term evaluation and up to a 1.8× reduction in total simulation time are achieved without compromising result accuracy. A more complex surrogate fuel is composed to match the liquid properties of commercial E10 gasoline. A complex gas phase mechanism with 8 fuel components and 382 chemical species representing E10 gasoline is then evaluated in the same engine, demonstrating that the acceleration achieved is maintained for more detailed gas phase kinetics. This capability enables broader use of detailed chemistry in ICE applications supporting modern powertrain development needs.