The focus of this dissertation is on the development and improvement of spray models for large–eddy simulation (LES) of turbulent two-phase flows in direct-injection spark-ignition (DISI) engines. The work can be regarded as a continuation of the development of LES framework at the Engine Research Center (ERC). The LES two-phase governing equations are solved using the Lagrangian-Eulerian (LE) approach in a variation of the OpenFOAM-2.3.x code developed by the OpenFOAM Foundation. A mixed-type one-equation dynamic structure turbulence model is used as the basis for turbulence modeling.
LES models are developed for DISI spray breakup, Sub-grid scale (SGS) turbulent dispersion, and SGS energy dissipation rate. The spray breakup model builds on top of the hybrid Kelvin-Helmholtz (KH)/Rayleigh-Taylor (RT) model by incorporating the bag/bag-stamen breakup regimes. A concept of RT breakup length is introduced to account for the plume-interactions and the effective nozzle diameter of DISI spray. The SGS models are developed in the context of LES and require the SGS kinetic energy, which is obtained by solving its transport equation in the turbulence model.
The performance of the new models is evaluated against a wide range of DISI spray experiments covering both early and late injection engine-like conditions. Examination of spray characteristics is performed for both global and local quantities such as penetration length, Sauter mean diameter (SMD), droplet velocity, liquid-phase concentrations, and spray envelop. The discussion focuses primarily on the DISI spray breakup, followed by a posteriori test results of the SGS models. An uncertainty quantification (UQ) study is also performed to analyze the impact of spray boundary conditions and breakup model parameters on LES of DISI sprays.
LES results show that the addition of models for bag/bag-stamen breakup regimes results in more accurate predictions of spray characteristics. The modified breakup length concept also predicts more realistic penetration curves across a range of ambient temperature and density conditions without tuning the model parameters. The improved SGS dispersion model correctly predicts local liquid-phase characteristics such as velocity and projected liquid volume fraction. A preliminary study of SGS dissipation rate modeling also shows that the SGS model is able to accurately predict the energy balance between the resolved and the SGS fields across various mesh resolutions.