Fuel spray impingement and liquid film formation in a gasoline direct-injection spark-ignition engine

Abstract

It is important to improve the thermal efficiency and reduce the harmful exhaust emissions of the direct-injection spark-ignition engine. However, this engine has problems such as the emission of particulate matter, including soot, from pool fire with luminous flames. Pool fire is caused by the thermal decomposition of a liquid film, which is created by fuel spray impinging on a piston surface. An understanding of liquid film formation process is important to reduce particulate matter. The purpose of this investigation was to evaluate the effects of injection pressure on fuel spray impingement and liquid film formation process, under engine motoring conditions, using the laser-induced fluorescence method. The fuel consisted of isooctane, 1-octanol and rhodamine B. 1-Octanol was the solvent for rhodamine B, which was illuminated with a neodymium-doped yttrium aluminum garnet laser, causing it to emit red fluorescence at a wavelength of 580 nm; the second harmonic of the laser is at 532 nm. Liquid film images were captured using a high-speed camera. Using image processing, the liquid film area, thickness and mass were estimated. It was found that increasing injection pressure increased the liquid film area, thinned the film and decreased the mass of fuel that remained. In total, 35.6% and 32.5% of the injection mass remained on the piston surface at an injection pressure of 5 and 13 MPa, respectively. In addition, the in-cylinder flow affected the liquid film formation process, stretching the film in the direction of the flow.

Graphical abstract

Appendix

A. Measurement uncertainty of liquid film thickness

The liquid film formation process could be captured and analyzed from below the piston with the RCEM. Good reliability of the measurement method and data analysis is necessary. However, when using the RCEM, the effects of in-cylinder pressure and temperature changes, in-cylinder flow and piston movement must be considered, which makes it difficult to evaluate reliability. In addition, using the RCEM, only the view from below the piston is available. For these reasons, the reliability of the experimental method and data was investigated out under atmospheric conditions.

Figure 10 shows a schematic diagram of the experimental setup under atmospheric conditions. Each component and the optical layout were the same in the motoring conditions in RCEM. Figure 11 shows the ${\delta }_{\mathrm{nfl}}$ (liquid film thickness) distribution at TDC timing for each value of P_inj. The liquid film thickness distribution was determined using the captured images and calibration equation as shown in Fig. 2. As P_inj increased, ${\delta }_{\mathrm{nfl}}$ decreased. Comparing the results for P_inj = 5 and 13 MPa, the maximum values of ${\delta }_{\mathrm{nfl}}$ were approximately 50 and 20 μm, respectively. By contrast, the minimum value of ${\delta }_{\mathrm{nfl}}$ was approximately 10 μm in both cases. It is assumed that $V$ affected the distribution; as P_inj increased, so did $V$; thus, during spray impingement, the increased $V$ pushed the liquid film over a wider and longer area. Therefore, the film became thinner and its area increased as P_inj increased.

Fig. 10

A schematic diagram of the experimental setup under atmospheric conditions

Fig. 11

The liquid film thickness distribution at TDC timing for each injection pressure under atmospheric conditions

Figure 12 shows the liquid film mass at TDC timing for each value of P_inj. For P_inj = 5 MPa, 0.82 mg formed on the piston surface, which was 18.0% of the total injected mass. By contrast, for P_inj = 13 MPa, 0.60 mg formed (13.2% of the injection mass). From the results of the side-view visualization, as P_inj increased, so too did the rebounded fuel. In addition, considering the breakup model, increasing P_inj increased the ${\mathrm{We}}_{n}$ of each fuel droplet, so they tended to splash rather than deposit. For these reasons, as P_inj increased, the liquid film mass decreased.

Fig. 12

The liquid film mass at TDC timing for each injection pressure

The purpose of this experiment was to evaluate the measurement method. The liquid film formation process was captured from below and beside the piston using a high-speed color camera. Comparing our results with relevant studies that used the LIF method (Schulz et al. 2015; Alonso et al. 2010), the ${\delta }_{\mathrm{nfl}}$ distributions matched and the maximum ${\delta }_{\mathrm{nfl}}$ values were almost the same. Therefore, it is considered the measurement method in this study to be sufficiently reliable and conducted a further experiment using the RCEM.