Effect of Subgrid Modeling on the In-Cylinder Unsteady Mixing Process in a Direct Injection Engine

Abstract

Fuel-air mixing in a direct injection spark ignition (DISI) engine occurs in a highly unsteady, turbulent and three-dimensional flow. As a result, any cycle-to-cycle unsteady variation in the mixing process can directly impact the performance of the DISI engine. To study the unsteady process in these engines, we have developed and implemented a large-eddy simulation (LES) approach with an innovative subgrid scalar mixing model based on the linear-eddy mixing (LEM) model into a commercial IC engine code (KIVA-3V). Time-averaged results of the simulations using the new LES version (KIVALES) are compared to the steady-state predictions of the original KIVA-3V. Significantly different in-cylinder turbulent fuel-air mixing is predicted by these two methods. Analysis shows that KIVALES resolves spatial features larger than the grid and that the subgrid kinetic energy adjusts to the LES resolution. As a result, KIVALES captures a highly unsteady, anisotropic fuel-air mixing process whereas a more diffused mixed field is predicted by the original KIVA-3V. This ability of KIVALES is attributed to the subgrid closure which scales the subgrid dissipation with the local grid size and thus, decreases the overall dissipation in the flow.