Experimental Studies Identifying and Minimizing Sources of Pollutant Emissions from Advanced Engines and Light Duty Vehicles

Abstract

To develop necessary strategies for achieving immediate-term emissions reductions, it is essential to characterize the performance of state-of-the-art powertrain hardware and evaluate methods that expand engine capabilities. This dissertation presents multiple approaches to minimize emissions at varying levels of system consideration. At the sub-system level, end-of-injection events for ultra-high pressure gasoline fuel injection (600, 900, and 1500 bar) that contribute to unburned hydrocarbon and particulate emissions in a gasoline direct injection engine were studied in a constant volume chamber using high-speed shadowgraph imaging techniques. Experiments were conducted at non-reacting and non-flash boiling conditions at injection pressure conditions which were not previously studied. Qualitative analysis of test results revealed that tip-wetting was highly dependent on injector nozzle geometry and occurred more frequently for the smaller nozzle orifice. Tip-wetting was also found to be insensitive to the injection and ambient chamber pressure values. Droplet size and count during fuel dribbling showed high correlation with increasing ambient pressure when varied from 1 to 20 bar. At the engine system level, a case study was performed to characterize the sensitivity of a genetic optimization algorithm (GOA) to its hyperparameters in the application of engine calibration. The study utilized a neural network surrogate of a production gasoline direct injection engine with the objective of minimizing brake specific fuel consumption for a given NOx limit. Results from 1225 trials showed that the GOA was highly sensitive to the number of genes selected, and mutation rate. The study showcased the potential loss in fuel economy and redundant computational efforts due to the selection of sub-optimal hyperparameter values. The findings from this study inform best practices and guidelines to develop GOA to achieve faster and more accurate engine calibration. At the vehicle system level, conditions leading to engine cold-start and the associated impact on vehicle emissions were characterized for the first time using a state-of-the-art plug-in hybrid electric vehicle (PHEV). Time-resolved emissions measurements and powertrain signals were analyzed on three drive cycle tests. The results confirmed that HPCS events were significant sources of NOx and particulate emissions, and the conditions leading to the HPCS varied significantly between the drive cycles. Specifically, demand for high vehicle speed and/or high traction power triggered engine cold-start events in the different drive cycles despite high battery state of charge. To further explore the phenomenon of HPCS on tailpipe emissions, additional hybrid architectures were studied in chassis dynamometer experiments. In the next study, tailpipe emissions and powertrain signals from three PHEVs were measured on the US06 drive cycles and analyzed. The results showed the strong influence of battery state-of-charge on the tailpipe emissions of the vehicle with the power-split architecture during cold-start. In addition, the power-split configuration allowed the engine to start at lower mechanical load, with significantly lower cold-start tailpipe emissions compared with the P2 type hybrids which experienced cold-start due to excessive traction power demand. Aggressive driving conditions produced the highest levels of criteria pollutants across all test vehicles. The combination of the outcomes of these studies provides quantitative demonstrations of opportunities to minimize engine-out emissions at conditions encountered during real-world driving. The study showcases how cold-start is the largest contributor to the cumulative emissions and how this phenomenon could be efficiently managed by integrating improvements at various systems involved in the vehicle powertrain to achieve significant reductions in emissions.