9.

4 Semipermeable Filter Media – Gas Pressure Filtration Without Gas Flow 325

Figure 9.32 Principle prerequisites for

Cake p1 > p2
semipermeable membranes.

Gas

p1

Membrane p2

Liquid

cases and eventually the cake cannot be desaturated properly anymore. Then,
the cake properties change from brittle to pasty and sticky. Although such
cakes cannot be desaturated well, the danger of shrinkage cracks exists and the
microporous membrane in any case makes sure that no gas breakthrough occurs
and the filtrate remains particle free. A roller discharge has proven as a safe
method to detach such thin and pasty particle layers from the filter medium.
The semipermeability of the filter medium can be regulated via its pore size
and wetting behavior because the capillary pressure depends on the pore size,
surface tension of the liquid, and wetting angle (cf. Figure 8.33 and Eq. (8.38)). To
withdraw liquid from the pore of the filter medium, a pressure difference must
be applied from outside, which is greater than the acting capillary pressure. If
the pores in the filter cake are greater than in the pores of the filter medium, the
capillary pressure in the cake is smaller than in the filter medium. If the filtration
pressure difference is chosen larger than the capillary pressure of the cake, but
smaller than the capillary pressure in the filter medium, the gas can penetrate the
cake and displace the pore liquid, but the filter medium remains fully saturated.
Figure 9.32 illustrates these interrelations schematically.
Rain-protective clothing uses this effect inversely. An extreme hydrophobic
polytetrafluoroethylene (PTFE)-membrane lets air and vapor pass through as the
wetting fluid, whereby the capillary pressure of the micropores is so high that the
impinging water droplets are not able to generate enough pressure to penetrate
the membrane. As result of these considerations follows that from the theoretical
point of view, a gasless filtration seems to be possible.

9.4.3 Realization of the Process in Lab and Pilot Scale

In the first step, suitable membranes are needed for the process. Suitable means
that they have at first to be hydrophilic and must exhibit a capillary entry pressure
or bubble point of at least 80 kPa for application on vacuum filters. Secondly,
they have to be mechanically stable enough for the tests in the pressure filter cell
and at the end on a rotary filter. For these purposes, hydrophilic membranes of
different pore size and material had been tested in the lab scale with particle-free
water regarding their capillary entry pressure. This pressure corresponds to the
largest pore with the smallest capillary pressure and determines the pressure, up
326 9 Selected Aspects of Filter Media for Cake Filtration

600
Water (γL = 72 mN m−1)
Capillary Polyamide
entry pressure
pcap,e (kPa)
Cellulose
acetate
400
γL γL
Cellulose d
300 ester δ

Acrylic polymer

200 PTFE
d = 2.0 μm

100
Δp = 80 kPa (vacuum filtration)
4 · γL · cos δ
d = 0.5 μm pcap =
d
0
0 1 2 3 4 5 6 7 8 9 10 11 12
(pore diameter)–1 d–1 (μm)–1

Figure 9.33 Membrane bubble point.

to which the membrane is able to hold back the gas completely. Water was used
because aqueous slurries were taken in the focus. In Figure 9.33, the capillary
entry pressure is plotted against the reciprocal value of pore diameter, given by
the membrane manufacturers.
As expected for each membrane material, a more or less linear relationship
between bubble point and inverse pore diameter could be observed. The different
slope of the straight lines can be interpreted as a result of differences in the
membrane wetting behavior and pore geometry. PTFE membranes, as typically
hydrophobic materials, were included after making them hydrophilic with
ethanol. Nevertheless, PTFE had shown the poorest results. From Figure 9.33, it
can be derived that for a safe gas sealing against an external pressure of 80 kPa,
pore sizes of less than 2 μm should be chosen for cellulose as best wetting
material and less than 0.5 μm for hydrophilic made PTFE. According to the
Young–Laplace equation for water, 25 ∘ C, complete wetting (𝛿 = 0), and 80 kPa,
theoretically a circular pore diameter of d = 3.6 μm would result.
In the next step, the question must be answered, whether the flow resistance
of principally well-suited membranes may eventually represent a constraint in
comparison to the filtration performance of conventional woven filter fabrics. For
this purpose, the filter medium resistance of membranes and filter fabrics with
different pore size were measured with particle-free water and various slurries.
The results of these measurements already are shown in Figure 6.4. It could be
demonstrated that there is of course a big difference between small flow rates
for microporous membranes of 0.2 μm pore size and woven fabrics with pore
sizes of more than 10 μm, if only pure particle-free water is filtered. If slurry is
 9.4 Semipermeable Filter Media – Gas Pressure Filtration Without Gas Flow 327

Cake height
hc (mm)

14 Hematite x50 = 26 μm/water

Δp = 200 kPa cv = 23.4% t1 = 14.3 s

12

0.2 μm-membrane 57 μm-

10 fabric

Flot.-coal x50 = 65 μm/water

8 0.2 μm-membrane Δp = 80 kPa cv = 30.0% t1 = 40.0 s

6
57 μm-fabric
Flot.-coal x50 = 65 μm/water
4
Δp = 180 kPa cv = 30.0% t1 = 20.0 s
Δp = 80 kPa cv = 30.0% t1 = 20.0 s
2 57 μm-
0.2 μm- fabric
memnrance
0
0 5 10 15 20 25
Number of experiments

Figure 9.34 Comparison of cake formation on membranes and fabrics.

separated, in most cases, the flow resistance of the filter media itself plays only a
subordinate role in comparison to the first bridge-building particle layer on the
filter medium surface, which has to be assigned additionally to the filter medium
resistance in practical cake filtration processes. The dominating flow resistance
is generated by the bottleneck of very small pores between the particles, which
are forming the bridges across the pores of the filter medium. If the micropores
of the membrane are smaller than those in the cake, the bottleneck is of course
located on the membrane itself, but after very short time, the absolute resistance
of the cake is much greater and the membrane resistance again can be neglected.
These results find their expression in the real cake formation performance,
which is documented exemplarily in Figure 9.34.
Membranes of 0.2 m pore size are compared with woven fabrics of 57 μm pore
size for different materials and different pressure differences. The cake height for
comparable conditions in all cases had been nearly the same. Some experiments
had been repeated several times without special membrane cleaning and no
spontaneous blockage of the membranes could be observed. These results are
an important prerequisite for an economical cake filtration with microporous
membranes because they represented in the investigated cases no constraint
regarding the filter throughput in comparison to conventional filtration.
Because the filtrate flow during the cake-deliquoring phase is significantly
smaller than during the cake formation, the membranes should also not hinder
this process step. However, the question arises whether the air, which migrates
toward the membrane and is stopped there, could have a negative impact on the
filtrate flow. Figure 9.35 answers this question.