Modeling of Turbulence, Combustion and Knock for Performance Prediction, Calibration and Design of a Turbocharged Spark Ignition Engine

In this thesis work, a downsized VVA Spark Ignition engine is numerically and experimentally studied. In particular, the following topics are considered: •In-cylinder turbulence and combustion processes; •Knock and cycle by cycle variation (CCV) phenomena; •Techniques aiming to mitigate knock occurrence and improve fuel economy such as EGR and water injection methods; •Intake system redesign to reduce the emitted gas-dynamic noise; •Engine calibration. A deep experimental campaign is carried out to characterize the engine behaviour. Indeed, engine system is investigated both in terms of the overall performance (torque, power, fuel consumption, air flow rate, boost pressure etc.) and of the intake gas-dynamic noise at full load operation. In addition, proper experimental analyses are peformed on the engine to characterize the CCV phenomenon and the knock occurrence. Measured data are post-processed to derive experimental parameters which syntetize CCV and knock levels, according to the engine operating conditions. A 1D CFD model of the whole engine is realized in GT-PowerTM environment. Refined “in-house developed” sub-models capable to reproduce turbulence, combustion, CCVs and knock processes are introduced into 1D code through user routines. First of all, the whole engine model is validated against the experimental data both in terms of overall performance parameters and ensemble averaged pressure cycles and intake gas-dynamic noise at part and full load operation. Cycle by cycle variation is reproduced through a proper correlation and consequently a representative faster than average in-cylinder pressure cycle is obtained. Then, the knock model, with reference to the latter pressure cycle, allows to evaluate a proper knock index and to identify the knock limited spark advance (KLSA), basing on the same threshold level adopted in experimental knock analysis. In this way, the knock model taking into account the CCV is validated at full load operation. Once validated, the original engine architecture is modified by virtually installing a “Low pressure” EGR system. 1D simulations accounting for various EGR rates and mixture leaning are performed at full load points, showing improvements in the fuel economy with the same knock intensity of the base engine configuration. Water injection technique is also investigated by virtually mounting a water injector in the intake runners for each engine cylinder. In a similar way, 1D analyses are carried out for various water/fuel and air-to-fuel ratios, highlightinig BSFC improvements at full load operation. Since the engine under study is characterized by higher intake gas-dynamic noise levels, a partial redesign of the intake system is properly identified and subsequently tested with 1D and 3D CFD simulations to numerically quantify the gains in terms of reduction in the gas-dynamic noise emitted at the intake mouth. Finally, a numerical methodology aiming to calibrate the considered engine at high load knock-limited and at part load operations is developed. First, it shows the capability to identify with satisfactory accuracy the experimentally advised engine calibration. In addition, it allows the comparison of different intake valve strategies, underlining, in certain engine operating conditions, the fuel consumption benefits of an early intake valve closure (EIVC) strategy with respect to a Full Lift one, due to a better combustion phasing and a reduced mixture over-fuelling. The developed automatic procedure presents the capability to realize a “virtual” engine calibration on completely theoretical basis and proves to be very helpful in reducing time and costs related to experimental activities at the test bench