• CFD
• combustion
• engine
• knock
• simulation

Knock is one of the most dangerous and limiting phenomenon for a spark-ignited internal combustion engine. A correct prediction of this event is crucial for the optimization of the engine design. CFD is a valid tool to investigate and take into account knock already during the early design phases, but consistent and accurate models and methods are needed for a proper identification of the phenomenon.
In this thesis work, carried out at the development centre of Porsche AG in Weissach in cooperation with GRUMO UniMORE, CFD research group of the Univerisità degli Studi di Modena e Reggio Emilia, a methodology for the prediction of knock through RANS CFD simulations is proposed. Furthermore, the analysis of hundreds of experimental data was made and both statistic variables and single cycle characteristics were considered and correlated with the CFD results.
Knock was evaluated using a tabulated auto-ignition delay time model. A correlation between CFD and experimental indexes is presented, showing the possibility of the prediction of knocking occurrence using RANS CFD simulations. The proposed correlation provides reliable results with many engine geometries and operating conditions, and, therefore, can be used in the early phases of the development, where very few or no experimental data are available. In the end, the mentioned knock model was tested with a sensitivity study, in order to evaluate its robustness and the influence of uncontrolled errors and uncertainty on the results.

Knock is one of the most dangerous and limiting phenomenon for a spark-ignited internal combustion engine. A correct prediction of this event is crucial for the optimization of the engine design. CFD is a valid tool to investigate and take into account knock already during the early design phases, but consistent and accurate models and methods are needed for a proper identification of the phenomenon. In this thesis work, carried out at the development centre of Porsche AG in Weissach in cooperation with GRUMO UniMORE, CFD research group of the Univerisità degli Studi di Modena e Reggio Emilia, a methodology for the prediction of knock through RANS CFD simulations is proposed. Furthermore, the analysis of hundreds of experimental data was made and both statistic variables and single cycle characteristics were considered and correlated with the CFD results. Knock was evaluated using a tabulated auto-ignition delay time model. A correlation between CFD and experimental indexes is presented, showing the possibility of the prediction of knocking occurrence using RANS CFD simulations. The proposed correlation provides reliable results with many engine geometries and operating conditions, and, therefore, can be used in the early phases of the development, where very few or no experimental data are available. In the end, the mentioned knock model was tested with a sensitivity study, in order to evaluate its robustness and the influence of uncontrolled errors and uncertainty on the results.