Spray Mixing in Engines

Abstract

The mixing of a fuel spray in an optically accessible direct-injection spark-ignition engine was investigated to yield both comparative conclusions between different engine characteristics and fundamental observations of mixing. Planar laser-induced fluorescence was used to acquire high-resolution fuel distribution images capable of visualizing the fine-scale mixing structure. The scalar dissipation, which describes the rate of fine-scale mixing, was found to be arranged in thin, sheet-like diffusion layers for all cases, similar to previously published results obtained in a turbulent jet. Furthermore, the cumulative distributions of the scalar dissipation rates were found to be self-similar for different engine flowfields, different injector types, and different fuels; and the self-similar state appeared to be exactly coincident with published results obtained in a turbulent jet. Finally, the measured layer thicknesses of the scalar dissipation rates were found to be similar to published measured layer thicknesses for the turbulent jet. Thus, the same fundamental processes govern the mixing of a direct-injection spray and a simple turbulent jet. Individual fuel distributions were characterized by two metrics calculated in a region of interest: the spatial variation, which was the second moment term of the scalar field, and the mean scalar dissipation, which was the average fine-scale mixing rate. Interestingly, the data-set averaged values of these two metrics plotted against one another for all the conditions tested (flowfield, injector, fuel injection timing, and fuel volatility) were correlated, forming a single characteristic curve extending from nearly homogeneous conditions with small values of both the spatial variation and mean scalar dissipation to highly stratified fuel distributions with large values of both the spatial variation and mean scalar dissipation. Because the slope of the characteristic curve changed depending on homogeneity, the mean scalar dissipation was a sensitive metric for highly stratified fuel distributions, while the spatial variation was a sensitive metric for nearly homogeneous fuel distributions. Comparative results were obtained for early and late injection operation regimes. With early injection (at 180� BTDC), the condition of the bulk flowfield was found to dominate mixing. Maximum swirl and maximum tumble flowfields (generated with a 240� shrouded valve) were found to generate more homogeneous fuel distributions with lower cyclic variability at the image time (20� BTDC) than a low-swirl flowfield (generated by the intake port geometry). An air-assist injector showed some improvement over a pressure-swirl injector with the low-swirl flowfield. Increased fuel pressure did not show improvement with the pressure-swirl injector. By completing experiments with two fuels with different boiling points (isooctane and pentane), vaporization was shown to play a minimal role in early injection mixing. The late injection experiments showed that the preparation of the bulk flowfield dominated the convection of the fuel cloud towards the spark plug. A constant amount of time was maintained between fuel injection and image for different bulk flowfields. Consequently, the maximum swirl flowfield convected the fuel cloud well past the spark plug, the maximum tumble flowfield placed a dispersed cloud in the field of view, and the low swirl flowfield placed a dense cloud in the field of view (the injection and image timings were determined with the low swirl flowfield from combustion experiments). Because the field of view was insufficient to analyze the full fuel cloud for several conditions, comparative results were limited. Statistics indicated that the maximum swirl and maximum tumble flowfields might have generated more homogeneous fuel distributions. An attempt was made to assess the role of flowfield turbulence on spray mixing using a valve deactivation mechanism. By deactivating the engine valves for a number of cycles before fuel injection, the turbulence generated by the intake process would dissipate. Combustion experiments illustrated the reduction of turbulence through poor flame propagation. It was discovered, however, that the bulk gas temperature also decreased. By using fuels with different boiling points and by changing the coolant temperature, it was shown that poor vaporization was significantly increasing fuel stratification. Still, an innovative gas jet experiment demonstrated that mixing was impeded when the flowfield turbulence was reduced, independent of vaporization issues. It appeared from the results of the early-injection experiments that mixing was highly dependent on a well-conditioned flowfield, while different injector characteristics showed minor effects. The late-injection results were inconclusive, but the condition of the bulk flowfield did affect the mixing.