Modelling and control of automotive coldstart hydrocarbon emissions

The problem of coldstart Hydrocarbon (HC) emissions from spark-ignition internal combustion engines is addressed in this dissertation. Up to 90% of HC pollution emitted by an automobile comes from the first 120 seconds of operation, known as the coldstart period.; A catalytic converter model is developed to predict the coldstart warmup behavior of the catalyst, accounting for heat transfer from the exhaust and to the surroundings. Heat generation of Carbon Monoxide (CO) and HC oxidation is taken into account and the catalyst conversion efficiency is parameterized using a 2-dimensional Wiebe function of air-fuel ratio (AFR) and temperature. Results show Oxygen storage of the catalyst can be neglected during coldstart conditions. Overall agreement of the model prediction of catalyst temperature with experimental data is very good.; A simplified combustion model of the engine is developed based on heat release analysis where the parameters of the burn profile are dependent on ignition timing and fuel flow rate. Intake manifold filling and fuel flow dynamics are added to the model to account for the dynamic development of in-cylinder AFR. Simulation results show the engine model behaves as expected with variations of the control inputs.; The desired engine exhaust properties are derived from the catalyst model. It is shown that the most important exhaust property to enable quick lightoff of the catalyst and reduced cumulative tailpipe HC emissions is the exhaust gas temperature. Variation of AFR, and hence HC, can provide additional warmup of the catalyst due to exothermic oxidation in the catalyst, thereby reducing overall HC emissions.; Tracking of desired exhaust gas temperature, AFR, and idle speed is maintained using sliding mode control laws. The engine is able to track the desired exhaust profiles with very good performance due to the relatively fast dynamics of the engine compared to the thermal dynamics of the catalytic converter.; The control algorithms developed take into account that AFR is available for measurement. Alternatively, AFR can be estimated from cylinder pressure using heat release analysis or crankshaft acceleration.