An investigation into multi-component gasoline biofuel droplet and spray using non-ideal vapour-liquid equilibrium models

Abstract

Fuel injection technology plays an important role throughout internal combustion engine development. The mixture preparation inside the combustion chamber affects the combustion process and the pollutant emissions. The formation of particulate matter is difficult to predict. Biofuels are introduced in both gasoline and Diesel engines to reduce combustion emissions. However, biofuels usually behave differently during the mixture preparation process. The thesis investigates the non-ideal phenomena on evaporation dynamics of multicomponent droplets and spray. The UNIFAC model has been implemented into the software to determine the activity coefficients that affect the evaporation process and mixture distribution inside the combustion chamber. The results demonstrate good agreement with existing measurement and numerical data. For the model validation, the E36 and E78 have been studied. Raoult’s law can reasonably approximate the evaporation process of the E36 mixture. However, the non-ideal Vapour-liquid equilibrium (VLE) model must be used to predict the evaporation of the E78 mixture properly. The evaporation dynamics of a four-component gasoline/ethanol spray was finally investigated. A ternary mixture surrogate composed of iso-octane, n-pentane, and n-decane has been used to provide a more realistic engine spray simulation. The Poynting factor is also introduced to give a more detailed non-ideal VLE model. The simulation results demonstrate that high ambient pressure significantly impacts the liquid evaporation process. The non-ideal evaporation model significantly affects the vapour composition during evaporation, which determines spray combustion and emission dynamics. Compared to other liquid activity coefficient models. The UNIFAC shows unique advantages for its simplicity. The model only requires properties of constituent functional groups of each liquid component. The requirement can be useful when there is no VLE data available for binary-component subsystems of the multicomponent mixture.