A Novel Chemical Clustering Approach for the Acceleration of 3D CFD Simulations of Internal Combustion Engines with Complex Fuels

Detailed kinetics simulations coupled with 3D CFD offer a powerful analysis tool for combustion and emissions. Such methods allow consistent modeling of multi-component fuels from evaporation to combustion and correctly capture the effects of local inhomogeneities created by preferential evaporation on the performance and emissions of modern powertrains. Such computations are extremely computationally demanding, prompting interest in the development of calculation acceleration techniques that can effectively balance the speed and accuracy of the chemical source calculation terms. Chemical kinetics clustering methods are widely used for that effect. However, such techniques must be not only effective but also robust with respect to the engine conditions and fuel composition changes, to reduce the computational demands introduced by the need to calibrate the parameters of the acceleration method itself. In this paper, an extended chemical kinetics clustering approach is proposed. A calibration methodology for the parameters of this acceleration method is then introduced, based on multi-point single time step optimization for a toluene reference fuel (TRF) surrogate with ESTECO modeFRONTIER, utilizing frozen 3D CFD fields obtained with the Realis Simulation VECTIS code. The optimal clustering parameters thus constructed are then fine-tuned through a DoE exercise performed in VECTIS for the combustion event with the TRF surrogate using a coarse computational mesh. The robustness of the optimal parameters is then evaluated through the application of different perturbations to the species fields. Finally, the optimal clustering parameters are applied to full-load simulations of a typical GDI engine with simple TRF and 8-species E10 gasoline surrogates. The results demonstrate that the acceleration parameters determined by this workflow deliver 2.4× to 4.2× acceleration of combustion source calculation and 1.3× to 1.8× acceleration of the overall simulation while preserving solution accuracy and keeping the resolution of NO and soot emissions within 10% and 15%, respectively. The proposed methodology facilitates the broader use of detailed chemistry in internal combustion engine (ICE) applications, supporting modern powertrain development needs.