Study of RP-3/n-butanol fuel spray characteristics and ANN prediction of spray tip penetration

The blending of aviation fuel with biomass is currently an effective way to meet the demand for sustainable aviation fuel. This study tested the physical parameters of fuel blends of 0∼30 % volume fraction n-butanol with RP-3 aviation kerosene, and the macroscopic characteristics of the spray penetration tip (STP), spray cone angle (SCA), and spray area (SA) of the blends were investigated by computational fluid dynamics at different injection pressures (100–160 MPa) and ambient pressures (0.8–1.5 MPa). A model using artificial neural networks (ANN) was built to forecast the STP of the fuel. The findings indicate that the blending of n-butanol has little impact on the spray characteristics of the fuel. The STP and SA of the blended fuel rise but have a small effect on the SCA as injection pressure increases. The STP and SA of the fuel gradually decrease, while the SCA becomes larger as ambient pressure increases. NET9 which has 6 input layers, 9 hidden layers and 1 output layer with the best performance (RMSE: 0.46213, R2: 0.99969) was selected and the prediction error of NET9 was less than 3 %. This indicates that the model has the best prediction ability and can be used in subsequent research. [Display omitted] •Combining CFD simulation with ANN to explore the spray characteristics of RP-3/n-butanol.•Physical properties and environmental conditions will affect the spray characteristics.•Optimizing the structure of ANN can improve the prediction accuracy of spray characteristics.•The BP neural network model has high applicability and the discrepancy is less than 3 %.