Fast NOx prediction methodology via 1D thermodynamical tools

Abstract

Accurate modelling of Nitrogen Oxide, soot, CO and UHC emissions from diesel engines plays a crucial role during the development phases of powertrain systems due to increasingly more strict emission legislation. Undoubtedly, generating accurate and robust methods of emission prediction will serve to global optimization of engine sys tems at very early stages of engine development. Engine component selection, accurate prediction of specific fuel consumption and defining the correct EGR strategy (low and mid-high) can only be achieved via reliable and fast NOx emission prediction. There are many possible ways of emission prediction in literature such as 3D, stochastic re actor, semi-empirical, phenomenological models and neural networks. However, these prediction methods either need excessive test data or simulation duration. On the other hand, using 1D simulation tools is a faster way of emission predic tion but has low accuracy. In this study, it is aimed to improve a fast and accurate NOx emission prediction methodology by utilizing 1D Models generated in GT-Suite c software. Two different heavy-duty diesel engines with two different combustion mod els are modelled and correlated to test data. A NOx emission prediction methodology is developed in 9L heavy-duty diesel engine model and experimented with the 12.7 L heavy-duty diesel engine model. In both studies, extended Zeldovich mechanism outputs included in the software is tuned via embedding calibration multiplier maps, depending on different engine operating parameters. Comparison of simulation results with the use of varying NOx calibration multiplier maps against test data, shows that the developed methodology can be used to predict NOx values with high speed and accuracy.