480 F. Zhao et al. / Progress in Energy and Combustion Science 25 (1999) 437–562 Fig. 40. Effect of the spray-tip velocity on the mean droplet size for a wide range of injectors and fuel rail pressures . was reported that for early injection the initial development of the spray is largely unaffected by the air motion due to the high spray momentum. Spray impingement on the cylinder wall was predicted to occur at 185 ATDC on intake, causing a rich mixture to remain near the piston crown throughout most of the compression stroke. At a crank angle of 25 BTDC on compression, the bulk tumble motion begins to decay rapidly and a fairly homogeneous stoichiometric mixture is produced. It was predicted that over 90% of the injected fuel is vaporized by 20 BTDC on compression. For the case of late injection, with the injection event initiated at 40 BTDC on compression, fuel impinges directly on the piston crown. It was found that only 50% of the injected fuel is vaporized by 20 BTDC on compression. This indicates that the particular geometric configuration that was analyzed would not be able to operate in the stratified-charge mode. Duclos et al.  studied the fuel–air mixing process of the Mitsubishi GDI engine using CFD. The computational results show that, for early injection operation, the homogeneity of the mixture is degraded by increasing the average equivalence ratio. For overall lean homogeneous operation the mixture is nearly homogeneous, while a rich region exists above the center of the piston for overall stoichiometric operation. The analysis for stratified operation shows that the fuel initially ignited by the spark plug comes directly from the injector and is not guided by the piston bowl. The main function of the bowl geometry would appear to be the confining of the fuel during the flame propagation process. It was noted that some improvements of the description of the interaction between hot surfaces and the spray (the wall-film sub-model) are required even for the early injection timing, as the fuel distribution in the cylinder directly affects flame propagation. Interactions between impinging liquid droplets and the piston are also an important consideration for stratified charge operation. The interaction of the fuel spray with the in-cylinder airflow was investigated by Kono et al. . Fig. 41(a) shows the combustion chamber geometry that was used, and illustrates the locations of the injector nozzle and spark plug. A single-hole nozzle was used to inject fuel into the piston bowl of the engine. Three spark plug locations were selected, all having the same distance from the nozzle tip. As a result, three injectiondirections were used to direct the fuel towards the spark plug, including one with swirl, which is denoted as the forward direction, one radial, denoted as the central injection, and one against the swirl, denoted as the reverse direction. The measured engine performance and emissions obtained using these three different injectiondirections with different quantities of injected fuel are shown in Fig. 41(b). The KIVA calculations of the spray-dispersion characteristics inside the piston bowl indicate that significant bowl wall wetting occurs for injection in the forward direction. A rich fuel zone appears in the vicinity of the cavity wall and the spark gap. The fuel is dispersed by the swirl, and a lean mixture zone is formed downstream of the injector. As a result, the measured fuel consumption and UBHC emissions were found to be higher for the forward direction than for the other injectiondirections. Large fluctuations in IMEP were noted, indicating that ignition and combustion are not stable for this injectiondirection. For the case of reverse injectiondirection, the tip of the spray is rapidly decelerated, and the mixture cloud is formed in the vicinity of the spark gap. As a consequence, good fuel economy and stable combustion are obtained. For the central injectiondirection, the spray development and penetration are not significantly influenced by the swirl, and the tip of the spray penetrates across the cylinder bowl and impinges on the far wall. Therefore, a combustible mixture is not formed around the spark gap. Interestingly, the UBHC emissions for central injection were found to be equal to that for the reverse injectiondirection for all of the fuel amounts that were evaluated. This work demonstrates the important relationship between the airflow field and the spray orientation.
F. Zhao et al. / Progress in Energy and Combustion Science 25 (1999) 437–562 481 Fig. 41. Effect of fuel injectiondirection on engine performance and emissions : (a) configuration of the cylinder head; and (b) measured engine performance and emissions (F: forward; R: reverse; C: central). The amount of fuel–wall impingement is known to vary significantly with the injection timing and engine speed, thus the injection timing must be optimized in order to avoid spray over-penetration and wall-wetting [20,40,161, 326]. There is a general consensus that the timing for early injection should be adjusted so that the spray-tip chases the receding piston without significantly impacting it. Fig. 42 shows an example comparison of the phasing of spray-tip penetration and the piston crown trajectory at 1000 rpm for various injection timings . The injection timing is critical to avoiding spray impingement for both early and late injection. To enhance the fuel evaporation and the fuel–air mixing process, it is necessary to set a minimum time interval between the end of fuel injection and the occurrence of the spark in order to avoid an over-rich mixture near the spark plug. As a result, the injection timing should be advanced as the engine speed increases. According to Matsushita et al. , the fuel injection timing must occur between the intake TDC and 160 BTDC on compression in order to provide enough time for the fuel to completely vaporize. For the avoidance of spray impingement, start-of-injection (SOI) timing is the most meaningful and for mixture preparation, end-of-injection (EOI) timing is the most applicable injection timing parameter. Both should be recorded during GDI engine development programs. Iiyama et al. [99,100] investigated the effect of injection Fig. 42. Spray-tip penetration and piston trajectory for an engine speed of 1000 rpm .