LES Modeling Study on Cycle-to-Cycle Variations in a DISI Engine 2020-01-0242
The reduction of cycle-to-cycle variations (CCV) is a prerequisite for the development and control of spark-ignition engines with increased efficiency and reduced engine-out emissions. To this end, Large-Eddy Simulations can improve the understanding of stochastic in-cylinder phenomena during the engine design process, if the employed modeling approach is sufficiently accurate. To assess the predictive capabilities of the turbulent combustion model used in this work, an engine-relevant Direct Numerical Simulation (DNS) dataset of premixed flame propagation in homogeneous isotropic turbulence is considered for a-posteriori investigations. LES predictions using the Flame Front / Progress Variable Equation Model are demonstrated to be in good agreement with the DNS results. Integral flame propagation results are shown to be unaffected by the choice of two eddy viscosity models, although some differences in the SFS velocity distributions near the flame front exist between the Dynamic Smagorinsky Model (DSM) and the Coherent Structure Model (CSM). The validated combustion model has been applied to investigate CCV in a direct-injected spark ignition (DISI) engine under fuel-lean conditions with respect to a stoichiometric baseline operating point. It is shown that the crank angle when a characteristic fuel mass fraction is burned, e.g. MFB50, correlates with the equivalence ratio computed as a local average in the vicinity of the spark plug. However, the lean operating point exhibits significant CCV, which are shown to be correlated with the in-cylinder subfilter-scale (SFS) kinetic energy.
Tobias Falkenstein, Marco Davidovic, Antonio Attili, Mathis Bode, Hongchao Chu, Seongwon Kang, Heinz Pitsch, Hiroyoshi Taniguchi