Quantitative Pollutant Modelling: an Essential Prerequisite for Diesel HCCI and LTC Engine Design

Abstract : To face the demand for efficient and environmentally-friendly engines, the Diesel HCCI (Homogeneous Charge Compression Ignition) and LTC (Low Temperature Combustion) concepts have been developed in order to drastically reduce the pollutant emissions of present Diesel engines at part load while maintaining their fuel consumption attractive. To be useful in the optimisation procedure of such new engine concepts, 3D CFD (Computational Fluid Dynamics) simulation softwares have to be predictive in terms of consumption and pollutants not only qualitatively (emphasising the global trends) but also quantitatively (providing reliable numbers to allow design or strategy comparisons). The implementation of accurate predictive combustion and pollutant models in 3D CFD codes are consequently a prerequisite to the use of these simulation tools to develop new combustion chamber designs or to test unconventional operating strategies. Pollutant modelling is a very difficult task because the pollutant formation and conversion during and after combustion heavily depend on the mixture formation and combustion processes that define the initial conditions for their complex chemical reactions. Furthermore, the occurring complex chemical reactions are far from being totally known. The correct prediction of the mixture formation and combustion processes are therefore essential for a correct pollutant prediction. However, the complex chemistry controlling the pollutant formation needs to be reduced to propose models that give accurate results with a reasonable CPU time consumption. For the HCCI and LTC engine concepts, the operating restrictions are mainly the NOx/soot trade-off and the noise level. The noise level may be related to the pressure rise and its prediction will therefore be a matter of accurate combustion description. On the other hand, the NOx/soot trade-off can only be numerically handled if accurate pollutant models are available. In the present study, the NOx model is the extended Zeldovitch model while the soot model is the PSK (Phenomenological Soot Kinetics) model coupled with the CORK (CO Reduced Kinetics) model to ensure a correct energy balance.