Integrating properties and conditions to predict spray performance of alternative aviation fuel by ANN model

Abstract

Alternative aviation fuel has been confirmed benefits for GHGs reduction and energy saving. Alternative fuel use should meet drop-in fuel requirement, and one of the important factors to ensure combustion completeness is to achieve spray requirement in the whole envelop of flight. Alternative fuels are characterized different fuel properties at low temperature comparison with traditional jet fuel. For understanding fuel properties and spray-related processes under different conditions, alternative aviation fuel, including Fischer Tropsch (FT), cellulose hydrotreating jet fuel (CHJ) and traditional jet fuel (RP-3), were investigated spray performance. According to empirical equation deduced from experiment data (283 K-343 K), deviations to RP-3 enhanced significantly on surface tension and viscosity at low temperature aera (243 K-273 K). As the complex and discontinuous interaction between nozzle structure and fuel properties with temperature, and thus it is difficult to obtain appropriate empirical equation or simulation results at low temperature. Moreover, non-drop-in fuel like pure FT fuel cannot comply with the same spray mechanism as drop-in fuel. The artificial neural network (ANN) approaches have been involved to solve the complex relationship of properties with spray performance. ANN-spray model coupling with ANN-mass flow can predict not only cone angle and liquid length but also SMD and velocity in liquid zone and droplet zone with above 0.99 total correlation coefficient. Coupling simulation results of mass flow and spray performance, FT and CHJ as well as blend fuels present more obvious difference to RP-3 in droplet size distribution and velocity distribution at low temperature.