Detailed Characterization of Negative Valve Overlap Chemistry by Photoionization Mass Spectroscopy 2015-01-1804

For next-generation engines that operate using low-temperature gasoline combustion (LTGC) modes, a major issue remains poor combustion stability at low-loads. Negative valve overlap (NVO) enables enhanced main combustion control through modified valve timings to retain combustion residuals along with a small fuel injection that partially reacts during the recompression. While the thermal effects of NVO fueling on main combustion are well understood, the chemical effects of NVO reactions are less certain, especially oxygen-deficient reactions where fuel pyrolysis dominates. To better understand NVO period chemistry details, comprehensive speciation of engine samples collected at the end of the NVO cycle was performed by photoionization mass spectroscopy (PIMS) using synchrotron generated vacuum-ultraviolet light. Two operating conditions were explored: 1) a fuel lean condition with a short NVO fuel injection and a relatively high amount of excess oxygen in the NVO cycle (7%), and 2) a fuel-rich condition with a longer NVO fuel injection and low amount of NVO-cycle excess oxygen (4%). Samples were collected by a custom dump-valve apparatus from a direct injection, single-cylinder, automotive research engine operating under low-load LTGC and fueled by either isooctane or an 88-octane research certification gasoline. Samples were stored in heated stainless steel cylinders and transported to the Lawrence Berkeley National Laboratory Advanced Light Source for analysis using a Sandia National Laboratories flame sampling apparatus.

For all isooctane fueled conditions, NVO cycle sample speciation from the PIMS measurements agreed well with previously reported GC sample measurements if the sum total of all isomer constituents from the PIMS measurements were considered. PIMS data, however, provides richer speciation information that is useful for validation of computational modeling approaches. The PIMS data also revealed that certain species for the GC diagnostic were either misidentified during the calibration process or not identified at all. Examples of unidentified species include several classes of oxygenates (e.g., ketenes, aldehydes, and simple alcohols) and simple aromatics (e.g., benzene and toluene). For the gasoline fueled NVO cycles, performance characteristics were well matched to corresponding isooctane fueled NVO cycles. However, significant PIMS cross-talk from a wide range of gasoline components restricted the sampling analysis to a handful of species. Nonetheless, it was confirmed that for fuel-lean NVO operation there was a comparable increase in acetylene with NVO injection timing retard that is attributed to the prevalence of locally-rich, piston-surface pool fires caused by fuel spray impingement.