Fuel spray modelling in direct-injection diesel and gasoline engines
Computational fluid dynamics methodologies have been achieving in the last decades remarkable progresses in predicting the complex physical process in internal combustion engines, which need to be continuously optimised to get the best compromise between fuel economy, emissions and power output/drivability. Among the variety of computational tools developed by researchers to investigate the multi-Phase flow development from high-pressure fuel injection systems for modem diesel and gasoline direct injection engines, the Eulerian-Lagrangian stochastic methodology, which models the air/vapour mixture as continuous phase and the liquid droplets as the dispersed one, has become standard among the developers of commercial or in-house university CFD codes due to its intuitive assumptions and simple implementation. It is generally recognised that this method is specifically suitable for dilute sprays, but it has shortcomings with respect to modelling of the dense sprays present in the crucial region close to the nozzle exit of fuel injection systems. Moreover, the mathematical formulation of the Eulerian-Lagrangian models is intrinsically related to critical numerical issues, like the difficulty of correctly estimating the initial conditions at the nozzle hole exit required by spray modelling calculations and, furthermore, the dependency of the results on the spatial and temporal discretisation schemes used to solve the governing flow equations. To overcome some of these difficulties, a modified Lagrangian methodology has been developed in this study. The interaction between the Eulerian and the Lagrangian phases is not treated on the cell-to-parcel basis, but using spatial distribution functions, which allow for distribution of the spray source terms on a number of cells located within a distance from the droplet centre. The end result is a numerical methodology which can handle numerical grids irrespective of the volume of the Lagrangian phase introduced. These improvements have been found to offer significant advances on Lagrangian spray calculations without the need to switch to Eulerian models in the near nozzle region. Besides these fundamental numerical issues, the present study offers some new insights on the physical processes involved in evaporating sprays under a wide range of operating conditions typical of advanced diesel and gasoline direct injection engines. Attention hag been directed on the topic of liquid droplet vaporisation modelling, which has been addressed by implementing and discussing different models published in the literature. Topics of particular emphasis include phase equilibrium, quasi-steadiness assumption, fuel composition, physical properties correlation, droplet shape and energy and mass transfer in the liquid and gas phases. The models have been implemented and validated against an extensive data base of experimental results for single and multi-component droplets vaporising under suband super-critical surrounding conditions and then implemented in the in-house GFS code, the multi-phase CFD solver developed within the research group over the last decade. A variety of physical sub-models have been assessed against comprehensive experimental data, which include the effect of thermodynamic, operating and physical parameters on the liquid and vapour penetration of diesel sprays. In particular, the effect of liquid atomisation, evaporation, aerodynamic drag, droplet secondary break-up and fuel physical properties has been thoroughly tested. The sensitivity of the predictions on the numerical treatment of the multi-phase interaction has been investigated by identifying and properly modelling the numerical parameters playing the most crucial role in the simulations. Finally the validated code has been used to investigate the flow processes from three high-pressure injection systems for direct injection spark-ignition engines. These have included the pressure swirl atomiser, the multi-hole injector and the outward-opening pintle nozzle. These investigations have enlightened the crucial role of the accurate modelling of the link between the internal nozzle flow prediction and the characteristics of the forming sprays in term of the successive multi-phase flow interaction, as function of the design of the fuel injection system used.