Stochastic combustion modeling of direct injection diesel engines

A comprehensive diesel engine combustion model has been developed. The model is an extension of similar thermodynamic multi-zone models for gasoline engines. In diesel engines a charge of air and residuals is present into which fuel is injected creating a highly inhomogeneous mixture and multi-phase spray structure. The non-uniformity on both local and global scales makes the diesel engine environment very difficult to characterize. In this model, the charge is divided into many distinct zones to express geometric information and to provide spatial resolution. Within each zone structure, mass, momentum and energy conservations are expressed as lumped formulations. This basic structure is coupled with a stochastic treatment of the properties and processes related to turbulent mixing and transport. This description plays an essential part in the modeling of the inhomogeneity and the turbulent mixing.; Besides the overall model development, the thesis improves the models for combustion product equilibrium composition and properties with an increased range of validity, which is crucial for diesel combustion simulations because of the wide local fuel-air ratio range. It also gives a robust solution procedure.; A new coalescence/dispersion model for stochastic turbulent mixings is proposed. The new model provides a more realistic description of the initial phase of mixing which strongly affects burn rate predictions.; Compressibility effects on turbulence are studied and significant improvements are made over the previous engine flow models. It is demonstrated that Reynolds stress models are needed to predict the non-equilibrium and anisotropic in-cylinder flows as they are subject to unsteady and nonisotropic compressions.; The overall model is demonstrated to correlate well with experiments with respect to performance. Pressure trace and heat release are shown as functions of major operating conditions: inlet pressure and temperature, fuel-air ratio, EGR, fuel injection conditions, turbulence and swirl levels. {dollar}NOsb x{dollar} and soot emissions are predicted as functions of load, EGR, injection timing and turbulence level. No comparison with experiments is made for emission predictions.