Journal of Chromatography Open (Nov 2022)

Why device design is crucial for membrane adsorbers

  • Franziska Hagemann,
  • Denis Wypysek,
  • Kristina Baitalow,
  • Patrick Adametz,
  • Volkmar Thom,
  • Matthias Wessling

Journal volume & issue
Vol. 2
p. 100029

Abstract

Read online

In preparative chromatography, a uniform flow distribution through the chromatography device is essential to get an effective ligand utilization and a breakthrough curve that is as steep as possible. Membrane adsorbers usually have a smaller aspect ratio (bed height to diameter ratio) than conventional columns. Upscaling devices with low aspect ratios is challenging due to the trade-off between easy-to-handle flow distribution and the pressure drop over the chromatographic membrane bed. This study aims to investigate the impact of membrane parameters (such as porosity and permeability) and device parameters (such as bed height, incident flow area, and dead volume) on the flow distribution by quantifying simulative and experimental residence time distributions (RTDs). In order to understand and develop a methodology to quantify the impact of the device geometry, the hydrodynamics of axial and radial devices are investigated using CFD simulations. The results show that the device configuration strongly influences the flow distribution. In contrast, the impact of the membrane characteristics on the flow distribution in the investigated small-scale devices with a cross-sectional area of up to 70mm2 is minor. However, membrane characteristics have a significant effect on the pressure drop over the stack. Experimental RTD data allow for evaluating only the impact of the whole device on signal responses. In this study, a purely model-based method was set up according to CFD simulations with the aim of separating the impact of membrane and membrane housing on the flow distribution. It has been shown that the investigated radial-flow devices have favorable hydrodynamic properties due to an optimized housing design and beneficial similar flow path lengths. Furthermore, an equivalent circuit of a CSTR and a DPFR was developed to predict the housing’s RTDs and, in the future, a model-based scale-up of devices.

Keywords