328 9 Selected Aspects of Filter Media for Cake Filtration

Moisture content
MC (wt%)
Flotation coal (x50,3 = 65 μm)/water
28 Δp = 80 kPa cv = 30%
Membrane (d = 0.2 μm)
Fabric(d = 57 μm)
26

Cake forming time

20 40

18

16

0
0 20 40 60 80 100 120
Deliquoring time t2 (s)

Figure 9.35 Comparison of cake deliquoring on membranes and fabrics.

It turns out that the deliquoring kinetics is behaving quite analogously for
both filter media. From these results on the one hand, the proof is made that
the microporous membrane is not impeding the deliquoring process, and on
the other hand, it is demonstrated that a gas flow through the filter cake is not
supporting the deliquoring. The only relevant mechanism to empty the pores is
the overcoming of the capillary pressure by the externally applied gas pressure
difference. This prerequisite has been made already previously in Section 8.3.4.2,
when the desaturation kinetics of filter cakes has been discussed.
After the principle proof of the process in the lab scale, comparative exper-
iments in pilot scale were tackled on a drum filter with a filter area of 0.7 m2 .
This drum filter was installed in a pressure vessel to enable not only vacuum fil-
ter experiments but also an operation of pressure differences up to 180 kPa. It
was accepted that the membranes had not been completely gas tight at the great
pressure, but it was interesting to see how the membrane would react on greater
pressure. For pressure differences of 80 kPa, the capillary entry pressure of the
membranes had not been overcome. In this connection, the difficulty appeared
that the available and in the lab scale with 20 cm2 easy-to-handle membranes did
not show sufficient mechanical stability in the pilot scale with the larger filter area
of 0.7 m2 and stronger mechanical impact during operation. As mentioned pre-
viously, enough robust polymeric filter media for such applications are not yet
available on the market. For this reason, it was necessary to fix membranes on
conventional filter fabrics in single manufacture and to generate by this measure
a filter medium, which exhibited both the desired microporous structure and the
required mechanical stability.
 9.4 Semipermeable Filter Media – Gas Pressure Filtration Without Gas Flow 329

Rel. cake height

0.8

n No membrane washing
0.6
α2
α1 Hemat. iron ore Flotation coal
0.4 x50 = 28 μm x50 = 260 μm
cv = 13.8 – 30.7 vol% cv = 10.0 vol%
Membrane Membrane
0.2 Drum filter
Polyamide, d = 0.8 μm Polyamide, d = 0.45 μm
Δp = 0.08 – 0.18 MPa
Fabric/weave Fabric/weave
n = 0.3 – 0.9 rpm Polyester/twill, d = 18 μm Polyester/twill, d = 18 μm
0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Drum revolutions r (–)

Figure 9.36 Comparison of cake formation for continuous filtration.

Figure 9.36 shows in the summarized form the results of the cake formation
for 2000 drum revolutions during a one-week campaign for two different slurries
and membrane filter media in comparison to a conventional woven filter fabric.
The operational parameters of the filter had varied several times during the
test campaign. During the shutdown times, the membranes were protected from
drying out to avoid crystallization of soluble substances in the membranes pores.
On the ordinate, the quotients of filtration with membrane filter cloth and con-
ventional filtration with pure filter fabric are drawn, and on the abscissa, the
number of drum revolutions realized during the test campaign. At a value of
1, the results of both filter media correspond exactly to each other. When the
results located below 1, the cake thickness on the membrane is comparatively
lower. For the flotation coal, the filtration results on the membrane got worse rel-
atively slowly but reached after a membrane cleaning again the original full value.
For the iron ore, the membrane blockage took place more quickly, and without
washing of the membrane after long time, a stable result of about 70% of the con-
ventional filter cloth was reached. From these results, it becomes clear that for the
application of microporous membranes in the field of cake filtration processes, of
course analogous phenomena can be observed, as well known from the cross-flow
filtration. Membranes have to be cleaned periodically to maintain their filtration
capacity. In the example, which is shown here, a combination of 250 μm mean par-
ticle size for coal and 0.45 μm membrane pore diameter is of course much more
favorable concerning the danger of membrane blockage than the combination of
28 μm mean particle size for iron oxide and 0.8 μm membrane pore diameter.
Beside the solid throughput, the accessible residual moisture content of the fil-
ter cake is interesting as process result. The results for the cake moisture content
during the test campaign, which is described above, are given in Figure 9.37.
The relative cake moisture content here again exhibits the value 1, if the
results of membrane filter cloth and conventional fabric are identical. If the
value becomes smaller than 1, the cake moisture for the membrane experiments
330 9 Selected Aspects of Filter Media for Cake Filtration

1.2
Total running time ttot = 65 h
Rel. cake moisture
MCrel (–)

1.0

0.9
n

0.8 α2
α1 Hemat. iron ore Floatation coal
x50 = 28 μm x50 = 260 μm
0.7 cv = 13.8 – 30.7 vol% cv = 10.0 vol%
Membrane Membrane
Drum filter
0.6 Polyamide, d = 0.8 μm Polyamide, d = 0.45 μm
Δp = 0.08 – 0.18 MPa Fabric/weave Fabric/weave
n = 0.3 – 0.9 rpm Polyester/twill, d = 18 μm Polyester/twill, d = 18 μm
0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Drum revolutions r (–)

Figure 9.37 Comparison of cake moisture content for continuous filtration.

is lower than that of the cake on the conventional fabric. This is mainly caused
by the different cake height. If the cake height is reduced, the moisture content
of the cake is lower for the same desaturation time. The slightly increased
membrane flow resistance due to some pore blockage is not the dominating
factor for the filtrate flow.
For the flotation coal, a curve resulted, which corresponded with the results
for cake formation. Identical moisture content resulted for identical cake height.
Slightly decreasing moisture content could be observed for increased membrane
resistance and thus decreasing cake height. Identical results were achieved again
after membrane cleaning at 1800 revolutions of the drum.
For iron oxide, the cake moisture content could be held more or less constant.
An interpretation for this phenomenon could be that here the increasing filter
media resistance in combination with decreasing cake thickness has led to a
compensation of negative and positive effects. There had been over the entire
operating time some cakes, which had lower cake moisture on the membrane
than on the conventional fabric. This may be caused by additional fluctuations
from the filter operation such as variations of the slurry concentration, eventu-
ally rewetting of the cake during cake discharge or similar effects. However, in
principle, the results obtained in the laboratory could be found and confirmed
in the pilot scale for continuous filtration. Nevertheless, beside the optimization
of the filter design, suitable membrane materials are needed in the future for
durable filter operation.

References
1 Smith, J. (2015) Assessment of ceramic filtration for a metallurgical process,
Dissertation, University of the Witwatersrand, Johannesburg.