Stochastic simulation of the spray formation assisted by a high pressure

Abstract

The stochastic model of spray formation in the vicinity of the injector and in the far‐field has been described and assessed by comparison with measurements in Diesel‐like conditions. In the proposed mesh‐free approach, the 3D configuration of continuous liquid core is simulated stochastically by ensemble of spatial trajectories of the specifically introduced stochastic particles. The parameters of the stochastic process are presumed from the physics of primary atomization. The spray formation model consists in computation of spatial distribution of the probability of finding the non‐fragmented liquid jet in the near‐to‐injector region. This model is combined with KIVA II computation of atomizing Diesel spray in two‐ways. First, simultaneously with the gas phase RANS computation, the ensemble of stochastic particles is tracking and the probability field of their positions is calculated, which is used for sampling of initial locations of primary blobs. Second, the velocity increment of the gas due to the liquid injection is computed from the mean volume fraction of the simulated liquid core. Two novelties are proposed in the secondary atomization modeling. The first one is due to unsteadiness of the injection velocity. When the injection velocity increment in time is decreasing, the supplementary breakup may be induced. Therefore the critical Weber number is based on such increment. Second, a new stochastic model of the secondary atomization is proposed, in which the intermittent turbulent stretching is taken into account as the main mechanism. The measurements reported by Arcoumanis et al. (time‐history of the mean axial centre‐line velocity of droplet, and of the centre‐line Sauter Mean Diameter), are compared with computations.