| Popis: |
Volvo Powertrain develops and produces heavy duty diesel engines for the various companies within the Volvo Group. In order to speed up the development process of new engines, computer simulation programs are used in order to predict system performance and emission formation. This thesis focuses on a computer model for Nitrogen Oxides (NOx) formation called VNOx. VNOx was originally developed by Volvo Technology together with the Austrian engine consultancy AVL in 1999. VNOx is a physical, zero- dimensional model that is based on the heat release and the evolution of the physical conditions within the cylinder. The objective of this thesis has been to calibrate the model so that predicted NOx levels match with measured NOx concentrations. In diesel engines the majority of NOx is formed due to the high temperature in the diesel flame, the chemical process that mostly predicts that formation is called the Zeldovich mechanism since it was first described by the Russian scientist Yakov Borisovich Zeldovich in 1946. NOx is a pollutant and is among other contributing to photochemical smog and acid rain. The project was divided into three significant steps: measurements, calibration and validation. The measurements were performed for two different engines, one utilizing Exhaust Gas Recirculation (EGR) and one that does not. The calibration consisted of tuning of the model constants, using a Matlab optimization tool. After the calibration the results were validated on a larger set of points. The validation showed good agreement for the engine without EGR, the standard error was on the order of 12 %. For the EGR engine the error was larger, but if an after-correlation was used the error could be reduced to roughly 15 %. The main reason for the error could be due to that the EGR fraction variation from cylinder to cylinder is not accurately taken into account. The analysis of the trends for the engine without EGR shows relatively good results, Also for the EGR engine the trends shows relatively good results except for the ability follow variations in injection pressure, mostly due to the limited variations in that parameter in the available data. The size of the standard error is on the same order as what have been presented by others for similar models. Validerat; 20101217 (root) |
