Modeling and Control of Raw Emissions of a Diesel Engine Under Practical Conditions

A real-time capable in-cylinder pressure based diesel engine-out PM estimator has been developed. Using the ECU signals and in-cylinder pressure data new variables have been derived and used as inputs for an exponential zero dimensional modeling approach. This approach required little computational effort making the ECU capable of cycle based PM emissions calculation, significantly faster than in real-time. Along with the PM estimator an accurate NOx emissions model has been utilized in a MIMO feedback controller motivated by the gain scheduling concept. Two types of experimental passenger car DI diesel engines, equipped with in-cylinder pressure sensors have been used. Measurements have been taken during steady state and transient operation on engine test benches for development work. Implementation of the emission models and the controller has been done on a test vehicle and tests were carried out on the test track and vehicle test bench. Good correlation between the estimated and measured PM has been achieved for various experiments, not only at steady state operation but also for transient states. Particularly, the model delivers good qualitative results in general, as well as good quantitative results in some regions. PM emission gradients between operating points are represented successfully. The raw emissions controller - despite further need for optimization - has been successful in controlling PM and NOx emissions simultaneously over EGR and pilot injection quantity. Gain scheduling has eliminated the need for an inverse combustion model. EGR and injector actuators were manipulated in a cascaded controller structure where the designed PI controllers altered reference values of the actual EGR and pilot injection quantity controllers that were already present in the system.