The increasingly stringent antipollution legislation and the necessity of a continuous control of the pollutant emissions lead to the development, among others, of NOx estimation models to be included in the engine control system and the on-board diagnostic system.
Two elements are important for a good estimation of these pollutant emissions: the knowledge of the combustion process, which is available via the instantaneous pressure signal, and the flame temperature, which is estimated from the in-cylinder conditions (mainly from the air temperature and oxygen concentration). All these parameters could be nowadays available and even analyzed on the vehicle during normal engine operation.
In this paper we intend to assess the NOx estimation sensitivity to inaccuracies in the input parameters to identify the critical parameters in this estimation.
The NOx emissions model which has been developed and used for this purpose in the frame of this research is based on the Zeldovich mechanism. As the NOx emissions prediction is highly dependent on temperature, a thermodynamic sub-model of the combustion chamber is required so as to estimate the adiabatic flame temperature. The input parameters for this model are the intake air mass flow rate, the injected fuel mass, the instantaneous in-cylinder pressure and the total in-cylinder mass (including both air and EGR), which traditionally is deduced from the volumetric efficiency and the intake temperature and pressure.
The sensitivity analysis shows that the accuracy of the in-cylinder air mass and the in-cylinder total mass is much more critical than that of the instantaneous incylinder pressure and the injected fuel mass for NOx prediction. For this reason special attention has to be paid to the intake air + EGR mixture temperature and fresh air mass measurements.