Comparison of two interfacial flow solvers: Specific case of a single droplet impacting onto a deep pool

A numerical study about a droplet impacting onto a deep pool is presented here by employing the Volume of Fluid method to track the interface. In this investigation, two open source solvers, named Gerris and Basilisk, are used to solve the incompressible Navier–Stokes equations with free surface. The results are compared to provide some reference for the researchers involved in this research community. Firstly, the capabilities of the two solvers in simulating the droplet impacting problems are validated against available experimental results. And the adaptive mesh refinement techniques used in the two solvers are also found to be applicable. When simulating the drop impacting problems in an axisymmetric coordinate system, it is found that less meshes are produced in Gerris than that in Basilisk when similar adaptive criterion is adopted in both solvers. However, Basilisk still shows much higher computational efficiency due to its great superiority at parallelization, and thus less CPU time is required for such unsteady problems. When simulating moderate Reynolds number impacting problems, the two numerical solvers show perfect agreement (Re≤5000), however, since higher Reynolds numbers (Re≥5000) require smaller size of mesh in vicinity of the thin impacting region and the concentrated vortex region, it is hard for Gerris to obtain satisfactory results because of its poor computational efficiency. Thus, with such a high parallel capability and computational efficiency, Basilisk is introduced to solve a large number of cases. It is summarized that the flow pattern of the drop impacting problem can be classified to three categories. They are smooth ejecta sheet, main vortex shedding and Von Kármán vortex street. Each flow pattern relates to the vortex evolution inside the liquid closely. Basililsk also shows its potential in solving much more complicated three-dimensional problems. They cannot be completed by Gerris owing to its poor computational efficiency.