Liquid fuel atomisation is largely affected by in and near nozzle phenomena. In this paper we establish a link between the initial droplet distribution at the nozzle exit, used as an input parameter in Lagrangian solvers, and the effect that this selection has on the subsequent predicted spray physics. Three initial PDFs with similar mean but different variance are tested in order to represent different aspects of the injection physics. Results are validated against an extensive range of experimental data from the Engine Combustion Network for the mildly cavitating Gasoline Spray G. Gasoline Direct injectors present high complexity both due to the importance of the near-nozzle phenomena for the fuel-air mixing, as well as due to the spray plume to plume interactions further downstream. Thus, they provide a well suited benchmark case for our study. We conclude that a correctly selected initial mean droplet size in combination to standard evaporation and break up models can provide reasonable predictions for a range of quantities including liquid and vapour penetration. Different droplet distribution models though exhibit differences in their prediction accuracy in terms of droplet dynamics depending on the initial droplet size variation around the mean. We demonstrate that these differences affect the coupling between the Lagrangian and the Eulerian dynamics through a Stokes numbers analysis close to the nozzle. We also analyse the effects that the stripping and catastrophic break up mechanisms have on the evolution of the droplet PDF at different axial locations depending on the selected initial droplet sizes. Overall, our results allow for a better understanding of the influence of the Lagrangian model input parameters on the global spray development on a numerical level, but can also help the optimisation of the design of new injectors by improving the understanding of the link between initial droplet distributions and the physics of the spray further downstream.
Read full abstract