(399u) The Influence of Regional Variation in Packing Density upon Productivity and Membrane Fouling

Key words:
hollow fibre membrane, direct-flow filtration, packing density, membrane
fouling

This
study focuses on operational considerations and design aspects of the direct-flow
filtration membrane module. Previously, optimal design ranges for various
geometrical parameters (packing density, fibre diameters, fibre lengths etc.)
were explored under assumption of packing uniformity and fibre dimensions in
unity. However, in manufacturing non-uniformity and irregularity of fibre spacing
cannot be avoided due to the process of manufacture and other possible causes such
as fibre misalignment and polydispersity of fibre diameter. Hence a model
assuming uniform packing is insufficient to describe the real fluid hydraulics
inside the shell of the hollow fibre module. Here we explore the effect of the
packing being essentially a mixture of tight and loose. Although such
maldistribution results in the channel fluid experiencing an element of
shell-side cross-flow this might not be disadvantageous. We are not aware of a
previous study on this effect. Our study is currently limited to axial flow and
unlike the design of GE there is no central collection tube.

For
a well-spaced packing hollow fibre module, an optimum packing density around 0.55
to 0.65 is preferred to ensure maximum module productivity. If an overall
packing density of 0.55 is divided into one loosely packed region and a tightened
one (Fig.1), the module will generate a higher pressure drop in the shell side
of the tightened region compared to that of the loose region (Fig.2). In
regions of very low local packing, the fibres are further apart and
interactions from neighbouring fibres are very small thus a larger amount of
fluid can pass through these regions. In Fig.2 a comparison is given between
the profiles for 0.55 packing density (black lines), 0.3 packing density (blue)
and 0.7167 packing density (red). The loosening from 0.55 to 0.3 has increased
the average TMP from 7.9 kPa to 8.4 kPa. However the tightening from 0.55 to
0.7167 has decreased the average TMP from 7.9 kPa to 5.7 kPa. Now the loose
region will only contain 21.8% of the filtration area so the overall effect is
a decrease in flux. For uniform packing the imposed flux was 23.7 LMH but for
the other arrangement shown in Fig.1 it would only be 18.8 LMH.

Additional
results that will be part of the presentation will include those on the
sensitivity of changes in fibre diameters, fibre lengths and imposed flux to
regional variation in packing density. 

Fig. 1. Regional variation of 40% area with 0.3 packing density, 60% area with 0.7167
packing density, and an equivalent module with a 0.55 packing density.

Fig. 2. Variation of fluid pressure in
the lumen side, p, and in the shell side, q with respect to regions in Fig.1 (Blue,
black and red lines are for packing densities of 0.3, 0.55 and 0.7167
respectively) at fixed pump pressure 10
kPa above permeate exit pressure (atmospheric pressure).