Optimization of a GDI engine operation in the absence of knocking through numerical 1D and 3D modeling

Various solutions are being proposed and adopted by manufactures and researchers to improve the en- ergetic and environmental performance of internal combustion engines within the transportation sector. For automotive spark ignition engines, gasoline direct injection is one of the presently preferred technolo- gies, in conjunction with turbocharging and downsizing. One of the limiting phenomena of this kind of engines, however, still remains the occurrence of knocking, namely the self-ignition of the so-called end- gas zones of the mixture, not yet reached by the flame front. This phenomenon causes strong in-cylinder pressure oscillations, high stress levels and even damage to engine components. Present work focuses on a numerical and experimental study of a turbocharged GDI engine and is aimed at assessing CFD-O (computational fluid dynamics optimization) procedures to be used in the phase of design as a decision making tool for the development of control strategies for a smooth and efficient operation. A preliminary experimental analysis is performed in order to characterize the con- sidered engine and to investigate the phenomenon of knocking that occurs under some circumstances as the spark advance is increased. The collected data are employed to elaborate a predictive criterion for the appearance of this kind of abnormal combustion, as well as to validate both a 1D and a 3D model for the simulation of the engine working cycle. Various numerical optimization procedures are then realized to increase the engine power output and simultaneously avoid conditions leading to undesired self-ignitions. These are either based on the use of a non-evolutionary algorithm or employ a genetic algorithm in the case multiple contrasting objectives are set. The response surface methodology is also explored as a way to reduce the computational effort.