Search

The size of drops generated by the capillary-driven disintegration of liquid ligaments plays a fundamental role in several important natural phenomena, ranging from heat and mass transfer at the ocean-atmosphere interface to pathogen transmission. The inherent non-linearity of the equations governing the ligament destabilization leads to significant differences in the resulting drop sizes, owing to small fluctuations in the myriad initial conditions. Previous experiments and simulations reveal a variety of drop size distributions, corresponding to competing underlying physical interpretations. Here, we perform numerical simulations of individual ligaments, the deterministic breakup of which is triggered by random initial surface corrugations. The simulations are grouped in a large ensemble, each corresponding to a random initial configuration. The resulting probability distributions reveal three stable drop sizes, generated via a sequence of two distinct stages of breakup. Four different distributions are tested, volume-based Poisson, Gaussian, Gamma and Log-Normal. Depending on the time, range of droplet sizes and criteria for success, each distribution has successes and failures. However the Log-Normal distribution roughly describes the data when fitting both the primary peak and the tail of the distribution while the number of droplets generated is the highest, while the Gamma and Log-Normal distributions perform equally well when fitting the tail. The study demonstrates a precisely controllable and reproducible framework, which can be employed to investigate the mechanisms responsible for the polydispersity of drop sizes found in complex fluid fragmentation scenarios.

We propose a novel method for the direct numerical simulation of interfacial flows involving large density contrasts, using a Volume-of-Fluid method. We employ the conservative formulation of the incompressible Navier-Stokes equations for immiscible fluids in order to ensure consistency between the discrete transport of mass and momentum in both fluids. This strategy is implemented on a uniform 3D Cartesian grid with a staggered configuration of primitive variables, wherein a geometrical reconstruction based mass advection is carried out on a grid twice as fine as that for the momentum. The implementation is in the spirit of Rudman (1998) [41], coupled with the extension of the direction-split time integration scheme of Weymouth & Yue (2010) [46] to that of conservative momentum transport. The resulting numerical method ensures discrete consistency between the mass and momentum propagation, while simultaneously enforcing conservative numerical transport to arbitrary levels of precision in 3D. We present several quantitative comparisons with benchmarks from existing literature in order to establish the accuracy of the method, and henceforth demonstrate its stability and robustness in the context of a complex turbulent interfacial flow configuration involving a falling raindrop in air.