CFD-Aided Optimization of Agitated Extraction Columns

Monday, October 17, 2011: 3:55 PM
202 A (Minneapolis Convention Center)
Enes Aksamija, Rainer Pfeffer and Matthäus Siebenhofer, Department of Chemical Engineering and Environmental Technology, Graz University of Technology, Graz, Austria

The applicability of Computational Fluid Dynamics (CFD) to extraction column research is constantly growing due to the incessant increase of calculation power. So far, the interfering phenomena are still far too complex to be predicted without using a bunch of models and assumptions. At present, CFD-simulations can provide hydrodynamic parameters (axial dispersion coefficient, bubble hold-up, droplet size distributions etc.) which are crucial for apparatus design.

In the present work, the continuous phase flow in a Rotating Disc Contactor (RDC150) was recorded using Particle Image Velocimetry (PIV). One could observe that the qualitative flow pattern did not vary under different operation conditions and appears to be a function solely of geometric dimensions. The mean velocity vector fields obtained from PIV were subsequently used to evaluate the RANS turbulence models. By performing simulations in Ansys Fluent 12.1, the Reynolds Stress Model turned out to give the best predictions.

The residence time distribution was obtained by injecting saturated salt solution and measuring the electric conductivity at four different axial positions. The obtained values for the axial dispersion coefficient were compared with Euler-Langrange Simulations (stochastic tracking method). The simulated values were systematically under predicted by CFD due to constructional differences between the lab column and the idealized 2D simulation domain (stator pins, bypass flow between stator discs and tower wall, etc.). Tendencies of the influence of volumetric flow, agitation speed and geometry on the axial dispersion were forecasted correctly. Based on the observation, that axial dispersion is almost independent of the dispersed phase, continuous phase simulations combined with the Euler-Langrange framework can be used to optimize the ratio of the column internals. The shaft diameter, the compartment height, the rotor diameter and the stator diameter were systematically altered in order to find the optimum compartment geometry that leads to lowest axial dispersion. The empirical design rules for the compartment geometry were evaluated and the missing correlation for the optimum shaft diameter was proposed using CFD.

Current research focuses on combining CFD with droplet population balances (DPB) to underline the optima found in the single-phase simulations. The Euler-Euler model was chosen as the multi-phase model and combined with the DPB add-on module provided by Ansys Fluent 12.1. An experimental drop size distribution (DSD) of Simon [1] (toluene in water) was divided into 29 drop size classes and set as boundary values at the inlet of the simulation domain. By implementing the breakage and coalescence kernels of Coulaloglou & Tavlarides [2] as user-defined-functions (UDF), excellent agreement with the experimental DSD of Simon [1] was observed. The geometric optima found in the single-phase simulations were subsequently investigated under dual-phase operation. Both higher hold-ups and narrower drop size distributions were observed for the optimized internals geometry, confirming the optimization in terms of overall column performance.

Since both single-phase and dual-phase (with DPB) simulations lead to comparable geometrical optima, mass transfer experiments with the optimized geometry are planned as a final proof for optimum overall design performance as well as validation of the simulation algorithm.



Simon, M. (2004). Koaleszenz von Tropfen und Tropfenschwärmen, PhD thesis, TU Kaiserslautern, Germany. [1]

Coulaloglou, C.A., Tavlarides, L.L. (1977). Description of interaction processes in agitated liquid-liquid dispersions. Chemical Engineering Science 32, 1289-1297. [2]

Extended Abstract: File Not Uploaded
See more of this Session: New Developments In Extractive Separations - II
See more of this Group/Topical: Separations Division