The purpose of this study is to use machine learning to build a model capable of predicting the flash boiling spray characteristics. In this study, the flash boiling spray was visualized using Shadowgraph visualization technology, and then the spray image was processed with MATLAB to obtain quantitative data of spray characteristics. The experimental conditions were used as input, and the spray characteristics were used as output to train the machine learning model. For the machine learning model, the XGB (extreme gradient boosting) algorithm was used. Finally, the performance of machine learning model was evaluated using R2 and RMSE (root mean square error). In order to have enough data to train the machine learning model, this study used 12 injectors with different design parameters, and set various fuel temperatures and ambient pressures, resulting in about 12,000 data. By comparing the performance of the model with different amounts of training data, it was found that the number of training data must reach at least 7,000 before the model can show optimal performance. The model showed different prediction performances for different spray characteristics. Compared with the upstream spray angle and the downstream spray angle, the model had the best prediction performance for the spray tip penetration. In addition, the prediction performance of the model showed a relatively poor trend in the initial stage of injection and the final stage of injection. The model performance is expired to be further enhanced by optimizing the hyper-parameters input into the model.

This study presents a prediction methodology of transport properties using the methane-based TRAPP (m-TRAPP) method in a wide range of temperature and pressure conditions including both subcritical and supercritical regions, in order to obtain thermo-physical properties for hydrocarbon aviation fuels and their products resulting from endothermic reactions. The viscosity and thermal conductivity are predicted in the temperature range from 300 to 1000 K and the pressure from 0.1 to 5.0 MPa, which includes all of the liquid, gas, and the supercitical regions of representative hydrocarbon fuels. The predicted values are compared with those data obtained from the NIST database. It was demonstrated that the m-TRAPP method can give reasonable predictions of both viscosity and thermal conductivity in the wide range of temperature and pressure conditions studied in this paper. However, there still exists large discrepancy between the current data and established values by NIST, especially for the liquid phase. Compared to the thermal conductivity predictions, the calculated viscosities are in better agreement with the NIST database. In order to consider a wide range of conditions, it is suggested to select an appropriate method through further comparison with another improved prediction methodologies of transport properties.

In this study, the thickness of the liquid sheet formed by a low speed impinging jet onto a flat plate was measured by the direct contact method. The spatial distribution characteristics of the sheet thickness in the radial and circumferential directions, and the effects of jet velocity and liquid viscosity were analyzed. The measurement results were compared with the theoretical predictions. The wavy surface was observed in the case of low viscosity water, but not in the high viscosity aqueous glycerol solutions. The sheet thickness increased as the circumferential angle increased or the distance from the impinging point increased, but the thickness decreased as the circumferential angle increased around the impinging point. As the jet speed increased, the sheet thickness decreased, and the sheet thickness increased as the liquid viscosity increased. Comparison with the theoretical predictions showed that the measurement results agreed well in the case of low viscosity water or high viscosity liquids around the impinging point. The distribution characteristics of the sheet thickness can provide useful means for prediction of spray characteristics in splash plate injectors.

To implement carbon-neutrality in transportation sectors until 2050, hydrogen is considered a promising fuel for internal combustion engines because hydrogen does not contain carbon itself. Although hydrogen does not emit CO2 emission from its combustion process, the low energy density in a volume unit hinders the adoption of hydrogen. Therefore, the understanding of hydrogen jet behavior and measurement of equivalence ratio must be conducted to completely implement the highpressure hydrogen direct injection. The main objective of this research is feasibility test of hydrogen local equivalence ratio measurement by laser-induced breakdown spectroscopy (LIBs). To visualize the macroscopic structure of hydrogen jet, highspeed schlieren imaging was conducted. Moreover, LIBs has been adopted to validate the feasibility of hydrogen local equivalence ratio measurement. The hydrogen injection pressure was varied from 4 MPa to 8 MPa and injected in a constant volume chamber where the ambient pressure was 0.5 MPa. The increased injection pressure extends the vertical penetration of hydrogen jet. Due to the higher momentum supply when the injection pressure is high, the hydrogen has easily diffused in all directions. As the laser trigger timing has delayed, the low hydrogen atomic emission was detected due to the longer mixture formation time. Based on equivalence ratio measurement results, LIBs could be applied as a methodology for hydrogen local equivalence ratio measurement.

Even in non-road UTV (Utility Terrain Vehicle), spark ignition engines are often used to reduce emissions. In this study, gasoline and LPG (Liquified Petroleum Gas) fuels were applied to UTV engines, and the exhaust gas and combustion stability were compared through engine tests. A 0.8-liter two-cylinder SI engine was used in the experiment. Experiments were conducted while changing the IVO (Intake Valve Open) and EVC (Exhaust Valve Close) at 1500 rpm 14 N·m, 40 N·m, and 3000 rpm 17 N·m, 44 N·m conditions. As a result of the experiment, when the valve overlap increased according to the change of IVO and EVC, combustion stability decreased and THC emission increased, but NOx decreased. Comparing the LPG engine with the gasoline engine, the amount of CO2 and PN (Particulate Number) generation decreased in the LPG engine, and the combustion stability was good.

Performance and emissions with different diesel injection methods were analyzed in a natural gas-diesel, dual-fuel engine under low-load conditions. Natural gas was supplied to intake port during the intake stoke to form a natural gas-air premixed mixture for all methods. Diesel was injected directly into the cylinder during the compression stroke in three ways: early injections, late injections, and a combination of early and late injections. The early injections had the highest thermal efficiency among the three methods owing to its highest combustion efficiency. The wide dispersion of diesel before the combustion initiation also allowed superior emissions characteristics.