Experimentation and Modeling of NOx Formation in a Small Turbo-Charged Diesel Engine

Abstract

In this thesis a detailed investigation into the primary formation characteristics and modeling approaches for the formation of NOx in turbo-charged diesel engines is presented. Through this investigation a multi-layered adiabatic core NOx model is developed and implemented into a single-zone internal engine combustion model. The comprehensive engine model includes estimation of heat release rates, heat transfer rates, cylinder pressure, cylinder temperature, and cylinder thermodynamic properties output on a crank angle basis. The NOx model investigation includes an analysis of different reaction mechanisms including the Extended Zeldovich Mechanism, N2O Mechanism, NO2 Mechanism, NH/HNO Mechanism, and variations/extensions of these mechanisms. Along with reaction mechanisms, other parameters that are analyzed include: temperature approximations, sensitivity to rate constants, validation of the quasi-steady state approximation, and the effect of multiple injections on simulation results. Common modeling approaches that were tested for their validity include fully mixed models, multi-layered approaches, and lumped modeling approaches. The developed NOx model is a semi-empirical model including: an empirical approximation for the formation of NO2 based upon operational characteristics, an exhaust gas recirculation model, and a combustion species equilibrium model. The EGR model is based upon burned gas composition properties which models charge dilution through mixture heat capacity and enthalpy and their respective effects on adiabatic flame temperature. The combustion equilibrium model is developed through the use of the minimization of Gibbs Free Energy and is similar to that of NASA’s CEA program. The equilibrium combustion model can produce species concentration of any specified species set with or without EGR, and includes model outputs such as mixture, enthalpy, entropy, frozen specific heat, and equilibrium specific heat. It was found through the development of the NOx model that the addition of an extra 43 reactions had a minimal effect on model accuracy. The rate constant analysis confirmed that the leading reaction in the Extended Zeldovich Mechanism is the most important reaction and serves as the rate limiting reaction for the mechanism. The quasi-steady-state assumption was validated showing only a 1~3% difference in comparison to a kinetically controlled reaction set. Pre-pilot injection scheduling is shown to have a minimal effect on simulation results with the majority of NOx formation originating from pilot and main injection scheduling. The resulting NOx model incorporates the sub-models described above coupled with the use of the Extended Zeldovich Mechanism, an approximated unburned air temperature between the ideal unburned air temperature and the mixture average, full injection scheduling, and a layered adiabatic core approach. The resulting model is shown to predict NOx concentrations, with and without EGR, to within 13% of experimental data. Additional model outputs include relative concentration predictions of carbon dioxide and oxygen based upon complete combustion principles.