Скачать с ютуб [PhD] Jacob Maarek : Mass Transfer at Fluid Interfaces at Large Péclet Numbers в хорошем качестве

[PhD] Jacob Maarek : Mass Transfer at Fluid Interfaces at Large Péclet Numbers

Abstract The characterization of mass transfer in unsteady multiphase flows is important for the understanding of various industrial and geophysical processes such as the desulfurization of steel in a metallurgical refining ladle or CO2 transfer across a wavy ocean surface. From a modeling perspective, the problem is challenging due to the formation of extremely thin, convection-dominated species boundary layers in the fluid which determines the global mass transfer rate. This dissertation, in the framework of the ERC program TRUFLOW, aims at enabling the quantitative prediction of mass transfer in such flows using simulation, high performance computation, and multiphysics, multiscale methods. The modeling of the multiphase flow is performed by direct simulation with a Volume-ofFluid based flow solver, using octree-based adaptive mesh resolution to treat the various hydrodynamic length scales present in the flow. The existing solver is improved for the case of modeling three-phase flows by the development and validation of a simple model for the computation of the capillary force near the three-phase contact line. A subgrid-scale (SGS) model for the treatment of interfacial thin species boundary layers is presented and integrated into the flow solver. Adapted from existing methods, the SGS model uses a non-linear reconstruction of the species distribution near the interface given by an analytic species boundary layer profile derived from first principles to correct convective and diffusive fluxes in the scalar transport equation. The proposed model is tested by characterizing species transfer from rising bubbles into a surrounding liquid, showing generally mesh-independent results of both steady-state and transient mass transfer rates at very large Peclet numbers (comparing convection and diffusion), given a well-resolved flow-field. Lastly, we generalize the developed SGS framework by substituting the analytic boundary layer profile with an approximation given by a shallow neural network, allowing one to simply and efficiently adapt the SGS model for various interface dynamics and transfer phenomena. The developed SGS model is applied to investigate mass transfer in a three-phase flow inspired by the ladle metallurgical secondary refinement process. A container of water, modelling the molten metal, is topped by a thin layer of oil, modelling the slag. The system is agitated by the injection of air at the bottom, creating a bubble plume that merges into the air at the top of the system. Thymol, acting as a passive scalar, is dissolved in the water and is progressively absorbed into the oil layer. We perform numerical studies of the experiment and validate results by comparison to previously performed physical experiments. The improved three-phase modeling scheme combined with the proposed SGS mass transfer framework leads to good agreement for global measures such as the open-eye size and mass transfer rate between the numerical simulations and the experiments. We next use local measures computed from the numerical experiments to build and validate a phenomenological model that explains the mass transfer rate as a function of the flow rate for the given experiment.