Tang, H., Burke, R., Akehurst, S., Brace, C. et al., "Behaviours of a GDI Gasoline Engine during Start," SAE Technical Paper 2014-01-1374, 2014, doi:10.4271/2014-01-1374.
Vehicle start-stop systems are becoming increasingly prevalent on internal combustion engine (ICE) because of the capability to reduce emissions and fuel consumption in a cost effective manner. Thus, the ICE undergoes far more starting events, therefore, the behaviour of ICE during start-up becomes critical.In order to simulate and optimise the engine start, Model in the Loop (MiL) simulation approach was selected. A proceduralised cranking test has been carried out on a 2.0-liter turbocharged, gasoline direct injection (GDI) engine to collect data. The engine behaviour in the first 15 seconds was split into eight different phases and studied.The engine controller and the combustion system were highly transient and interactive. Thus, a controller model that can set accurate boundary conditions is needed. The relevant control functions of throttle opening and spark timing have been implemented in Matlab/Simulink to simulate the behaviours of the controller. Good agreements were found between the measured and predicted control parameters. The accuracy of the engine speed input to the controller model was crucial to the prediction of the control parameters. Therefore, the engine friction and the combustion model were essential to the overall accuracy of the MiL simulation.The combustion during engine start has been analysed. Cyclic variability resulted in large variation in the 10-90% burn duration during start. Difference of up to approximately 50 degrees crank angle was observed between two adjacent combustion cycles under same operating condition. This highlights the necessity of simulating the combustion cyclic variability as the Coefficient of Variation in IMEP (COVIMEP) during start is an important factor to optimise. In addition, with the capability of modelling the interaction between the controller and the combustion system, advanced control strategies can be assessed in the co-simulation environment.