In this paper the numerical method Smoothed Particle Hydrodynamics (SPH) is assessed with the objective to estimate its suitability for predicting multiphase flows as they occur in fuel atomizers of aircraft engines. Discussed criteria are aspects of the physical modeling as well as computing issues. The modeling requirements are a proper multiphase flow handling, the correct prediction of surface tension effects and appropriate boundary conditions. As primary atomization implies length and time scales which cover several orders of magnitude, the numerical method has to provide good efficiency and scalability.
In order to evaluate the capabilities of the new method, simulations of a generic airblast atomizer have been performed using the well established Volume of Fluid (VoF) method and SPH. The study covers quantitative comparisons of flow fields, predicted fluid structures and droplet sizes. Furthermore, the parallel efficiency and the runtime of both methods are compared. It is found that SPH and VoF yield similar results in terms of physical correctness but differ dramatically regarding the computing behavior. The grid based method shows superior efficiency for a low number of CPUs. The SPH method scales much better at a high number of CPUs and reveals its potential for handling realistic atomizer configurations comprising a very large number of particles.