LES of spray and combustion in an internal combustion engine

With the ever rising need for better fuel efficiency and lower emissions, the development of better engine technology is essential. Different strategies are being considered to increase engine efficiency, such as direct injection or downsizing. Certain regimes are clearly identified as optimal for certain range of loads. For gasoline engines, the Homogeneous Charge Compression Ignition (HCCI) concept produces very low pollutants for low loads. At higher loads, however, the fast pressure rise can lead to engine damage and high noise levels. The Spark Ignition (SI) regime is the best suited for those cases. But pollutants emission is very high. In order to combine the best of the two regimes, a strategy consists in developing a hybrid SI and HCCI engine, functioning in HCCI regime at low loads, and SI regime at high loads. As the complexity of these strategies increases, the issue of monitoring and controlling the regime transition becomes a challenge. Therefore, numerical simulation is becoming an increasingly necessary complement to experiments. In order to make a significant impact simulation needs to reach two purposes. First, in the turbulent reactive flows occurring in Internal Combustion (IC) engines, the mixing of the reactants enabling the chemical reaction happens at the turbulence scales. Capturing this process accurately is essential to predict fuel and air distribution in the cylinder. Secondly, describing the dynamics of the transition between the two modes requires to capture cycle-to-cycle variations. With Reynolds Averaged Naviers-Stokes (RANS) approaches, all turbulence scales are modeled. Hence it does not accurately predict turbulent mixing. Besides RANS methods loose information about various unsteady processes, making it unable to predict cycle-to-cycle variations. With the increase in computational ressources, Large Eddy Simulations (LES) can now be considered. In contrast to Reynolds Averaged Navier-Stokes (RANS), LES can capture cyclic variations, which is essential when transitioning from one regime to another. Turbulence mixing is also better captured. Simulation of IC engines requires to solve for multi-physics processes: gas exchange, spray injection and combustion. This work investigates the spray modeling in this context. A Lagrangian method is used for this purpose. Even though spray models already exist in RANS, they cannot be used as are. Indeed, spray evaporation is a key process conditioning the structure of the mixture. Many models are proposed in the literature, so the first step is to determine the evaporation model best suited to our case. Spray models contain several phenomenological parameters that need …