Modelıng of NOx Emmıssıons in Internal Combustıon Engıne

Abstract

In order to reduce fossil fuel emissions, which are among the biggest threats of human health in the global world, regulations are getting harder every day. Many technologies continue to be developed to reduce/eliminate emissions during combustion (at the source) or after combustion. Instead of physical sensors used for the measurement of emissions, virtual sensors (VS) are developed and used by physical, gray box and black box modeling methods. In the current study, the NOx emissions generated at the engine output are modeled with steady state data by semi-physical modeling technique. The studies were carried out with data from 13 l, 6 cylinder and 353 kW diesel engine. The physical value used for the model is the maximum temperature calculated via a temperature sub-model based on the in-cylinder pressure measurement. The input data of the model are; calculated engine power at the electronic control unit (ECU), the estimated exhaust gas recirculation (EGR) ratio from ECU, the fuel rail pressure , the amount of injected fuel per stroke and the amount of trapped in cylinder air mass. The NOX model created by these parameters are converted into empiric formula by statistical methods. The engine NOX sensor data used as the target output data for validating and training the model parameters. Then validation results showed that the prediction accuracy of the developed NOX model is between 1% and 10% range that can be acceptable for internal combustion applications.