Christoph Kratzsch

Assessment of particle-tracking models for dispersed particle-laden flows implemented in OpenFOAM and ANSYS FLUENT

ABSTRACT In the present study two benchmark problems for turbulent dispersed particle-laden flow are investigated with computational fluid dynamics (CFD). How the CFD programs OpenFOAM and ANSYS FLUENT model these flows is tested and compared. The numerical results obtained with Lagrangian–Eulerian (LE) point-particle (PP) models for Reynolds-averaged Navier–Stokes (RANS) simulations of the fluid flow in steady state and transient modes are compared with the experimental data available in the literature. The effect of the dispersion model on the particle motion is investigated in particular, as well as the order of coupling between the continuous carrier phase and the dispersed phase. First, a backward-facing step (BFS) case is validated. As a second case, the confined bluff body (CBB) is used. The simulated fluid flows correspond well with the experimental data for both test cases. The results for the dispersed solid phase reveal a good accordance between the simulation results and the experiments. It seems that particle dispersion is slightly under-predicted when ANSYS FLUENT is used, whereas the applied solver in OpenFOAM overestimates the dispersion somewhat. Only minor differences between the coupling schemes are detected due to the low volume fractions and mass loadings that are investigated. In the BFS test case the importance of the spatial dimension of the numerical model is demonstrated. Even if it is reasonable to assume a two-dimensional fluid flow structure, it is crucial to simulate the turbulent particle-laden flow with a three-dimensional model since the turbulent dispersion of the particles is three-dimensional.

Numerical study of effects of pour box design on tundish flow characteristics

Abstract In the paper, the effects of special pour box design on tundish flow characteristics are investigated with the help of computational fluid dynamics. The numerical model is based on an Eulerian–Lagrangian approach for melt flow and non-metallic inclusion behaviour, respectively. For a prototypical twin strand tundish configuration, basic features of the mean tundish flow, residence time distributions and parameters of the melt history are discussed. It is shown that the pour box design is not only of importance for the tundish flow field but also of significance for processes of secondary metallurgy.

Influence of drag closures and inlet conditions on bubble dynamics and flow behavior inside a bubble column

ABSTRACT In this paper, the hydrodynamics of a bubble column is investigated numerically using the discrete bubble model, which tracks the dispersed bubbles individually in a liquid column. The discrete bubble model is combined with the volume of fluid approach to account for a proper free surface boundary condition at the liquid–gas interface. This improves describing the backflow region, which takes place close to the wall region. The numerical simulation is conducted by means of the open source computational fluid dynamics library OpenFOAM®. In order to validate the numerical model, experimental results of a bubble column are used. The numerical prediction shows an overall good agreement compared to the experimental data. The effect of injection conditions and the influence of the drag closures on bubble dynamics are investigated in the current paper. Here, the significant effect of injection boundary conditions on bubble dynamics and flow velocity in the studied cavity is revealed. Moreover, the impact of the choice of the drag closure on the liquid velocity field and on bubble behavior is indicated by comparing three drag closures derived from former studies.

Mathematical Modeling of Molten Salt Electrolytic Cells for Sodium and Lithium Production

Sodium (Na) and Lithium (Li) are produced using molten salt electrolysis. The electrochemistry of the electrolyte is well-researched; however, there are benefits to understanding the melt flow and implications on it for cell design modifications. The basic configuration of alkali metal cells is the Downs cell. This consists of a central anode surrounded by a cathode, and this geometry was the basis for this mathematical modeling study. The behavior of gas bubbles in molten electrolyte was studied in both Na and Li cells through the use of computational fluid dynamics (CFD) techniques. The distance between the anode and the cathode was varied in the CFD model to ascertain whether strong circulatory flows would change significantly in the cell. The standard k-e turbulence model was used to mimic turbulent flow, and a two-way coupled Discrete Phase Model (DPM) was adopted to simulate flotation behavior of chlorine bubbles and liquid metal droplets. The liquid metal distribution on the free surface was predicted using the Volume of Fluid (VOF) multi-phase model.

CFD of the MHD Mold Flow by Means of Hybrid LES/RANS Turbulence Modeling

Abstract In the last decades, electromagnetic braking (EMBr) systems become a powerful tool to dampen possible jet oscillations in the continuous casting mold. Further studies showed that if a EMBr is not positioned correctly, it can induce flow oscillations. Hence, the design of these braking systems can be promoted by adequate CFD simulations. In most cases, unsteady RANS simulations (URANS) are sufficient to resolve low-frequency, large-scale oscillations of these MHD flows. Alternatively, Large Eddy Simulations (LES) may also resolve important details of the turbulence. However, since they require much finer computational grids, the computational costs are much higher. A bridge between both approaches are hybrid methods like the Scale Adaptive Simulation (SAS). In this study, we compare the performance of SAS with URANS and LES. Results are validated in detail by comparison with data from a Ruler-EMBr model experiment.

Comparison of different Methods to model Transient Turbulent Magnetohydrodynamic Flow in Continuous Casting Molds

Modeling of the processes in the continuous casting mold engaged many scientists once the computer-technology was able to accomplish that task. Despite that, CFD modeling of the fluid flow is still challenging. The methods allow deeper and deeper inside views into transient flow processes. Mostly two kinds of methods are applied for this purpose. URANS simulations are used for a coarse overview of the transient behavior on scales determined by the big rollers inside the mold. Besides, LES were done to study the processes on smaller scales. Unfortunately, the effort to set up a LES is orders of magnitude higher in time and space compared to URANS. Often, the flow determining processes take place in small areas inside the flow domain. Hence, scale resolving methods (SRS) came up, which resolve the turbulence in some amount in these regions, whereas they go back to URANS in the regions of less importance. It becomes more complex when dealing with magnetic fields in terms of EMBr devices. The impact of electro magnetically forces changes the flow structure remarkably. Many important effects occur, e.g. MHD turbulence, which are attributable to processes on large turbulent scales. To understand the underlying phenomena in detail, SRS allows a good inside view by resolving these processes partially. This study compares two of these methods, namely the Scale Adaptive Simulation (SAS) and the Delayed Detached Eddy Simulation (DDES), with respect to rendition of the results known from experiments and URANS simulation. The results show, that the SAS as well as the DDES are able to deliver good results with higher mesh resolutions in important regions in the flow domain