Note: Descriptions are shown in the official language in which they were submitted.
-- 1 --
CLEANING OE~ F:LI.TERS
This invention relates to the cleaning of filters
and more particularly to the cleaning of filters made of
hollow fibres having a pore siæe in the range from 5 micron
to 0.01 micron.
Membrane fouling is generally recognised as the
outstanding problem in modern ultrafiltration. A full
discussion of the problems can be found in "Fifteen Years of
Ultrafiltration" by Michaels, A.S. in Ultrafiltration
Membranes and Applications edited by A.R. Cooper (American
Chemical Society Symposium, Washington, 9-14 September,
1979, Plenum Press, New York (1980); ISBN 0-306-40548-2)
where it is stated:
"the problems of reduced throughput
capacity, increased power consumption,
compromised separation cap~bility, and
reduced membrane service lifetime
associated with macro-, solute~ and
colloid-fouling of ultrafiltration
membranes have stubbornly resisted
adequate solution despite ten years of
engineering experience in pilot- and
full-scale industrial situations."
According to Michaels, back-washing by reverse
flow of permeate in hollow-fibre membrane modules,
significantly aids unplugging of mem~rane pores and
detachment of adhering deposits. However, there are only
two specific examples of permeate back-washing described in
this text and these concern filtration of town water and of
electro-deposition paints emulsified in water.
As set forth at pages 109 to 127 of the above
text, back-washing of hollow flbres with permeate is used
where operating transmembrane pressures are only about one
, - ~;,.;
', -
.
: '
..
~ 5~t;~
atmosphere so that particles are not driven hard into
membrane pores during the filtering process. As indicated
above, permeate back-washing has been used where the fouling
species are in liquid paint emulsion droplets as these
species do not wedge into the membrane pores as do solids.
~s the transmembrane flux is often only five to twenty
litres per square metre per hour (L/m2 hr), the
corresponding fluid velocity is only a few millimetres per
hour and there is, therefore, no possibility of a high
velocity cleaning action.
Permeate back-washing is, in essence, a recycling
process and thus a sacrifice of production rate is only
justified when the cleaning effect is significant. Some
sticky natural wastes (such as brewing residues, starch, and
egg~ are not removed to any appreciable extent by permeate
back-washing. Permeate back~washing is, by definition, a
purely hydraulic flow through totally wetted pores of the
ultrafiltration membrane. ~ollow, porous fibre ultrafilters
are preferred where back-wash cleaning is needed because of
the structure of the hollow fibres.
It is an object of this invention to provide an
improved method of back-washing hollow fibre filters which
uses gas as the back-wash medium.
The penetration of gas into the pores of a
membrane is resisted by the surface tension forces of the
contained wall-wetting liquid according to well known
theory. Indeed, surface tension is conveniently measured by
the breakthrough pressure needed to force a bubble out of a
submerged orifice. For common systems (such as oil in
hydrophobic pores or water in hydrophilic pores) the
breakthrough pressure required ranges from ten kilopascals
to a thousand kilopascals. The breakthrough pressures are
much higher than the usual operating pressures of the
~ilter.
75~ti()
-- 3
Prior art hollow-fibre type ultrafilters are usually
fed from the inside of -the fibres for many well known reasons.
However, according to the present invention, feedstock is applied
to the outside of the fibres and gas is introduced into the lumen
of the fibre as the back-wash medium. In some cases, the lumen
pressure swells a suitably designed fibre so that the pores are
enlarged whereby the particles are freed and swept away in the
expansion of the back wash gas.
The products of our co-pending International Patent
Application PCT/AU84/00179 (published September 12, 1984)
"Treatment of Porous Membranes" are ideal for gas back-wash since
they are highly elas-tic hydrophobic, relatively coarser porous
membranes which have a tenacious hydrophilic coating and
interstitial hydrophilic packing.
The packing prevents por~ collapse when fed under
pressure from outside the porous tube. The packing is of
different resilience and allows controlled expansion of the
(formerly hydrophobic) pores to which it adheres. It is thus
possible to design the composite fibre so that the gas release
characteristics are ideal for cleaning off different types of
blockage in differing configurations of fibre bundles without
pressures greater than those needed to sweep away the deposits
rapidly.
In some cases, especially where very fine-pored
interstitial material is deposited in relatively coarse-pored
base fibre, it is advantageous to back-wash first with a small
amount of permeate already in the membrane lumen and follow with
the high pressure gas back-wash. In this way, the small amount
of permeate adequately washes out fine blocking material from
within the interstices, and they overall cleaning is completed by
the higher pressure gas swelling the basa pores and erupting
around elastic openings. The pores must close again rapidly to
reseal the holes and the base material must not crack by work
hardening and must remain within its modified elastic limit,
.~
Polypropylene base is very resistant to flex
cracking but is hydrophobic and rather too ea~y to cru~h.
Its properties may be improved by elastic ~odification which
can be accomplished by the method of our above-mentioned co-
pending International patent application. The twoinventions together synergistically improve the performance
of microfilters and ultrafilters.
The use of gas as a back-wash medium enables the
removal of fouling species by explosive decompression of the
gas through the me~brane structure for the minor part and at
the outer membrane surface for the major part. Thus, it is
preferred that the gaseous back-wash step is carried out at
a pressure which is sufficient to overcome the effect of the
surface tension of the continuous phase of the feedstock
within the pores of the membrane.
In order that the invention may be more readily
understood and put into practical effect, reference will now
be made to the accompanying drawings in which:
Fig. 1 shows the flux profile for an illustrative
example of a gaseous back-wash system
incorporating a partial permeate back-wash,
and,
Fig. 2 is a schematic diagram of a gaseous back-
wash cleaning system according to one
embodiment of the invention.
In Fig. 1, OA represents initial permeate flux, OC
the filtering time and GH the recovered permeate flux. The
length CF represents the time of permeate back-wash and FG
the time of gas back-wash. For a given set of operating
conditions, the area ABCO depends on the rate of flux
decline and length of time between successive back-wash
operations. The area CDEF represents the volume of permeate
back-wash.
35In order to obtain optimal throughput, it is
necessary to simultaneously:-
. ', ~ , '. ' . : ,
. ' ',' :
.
~L~'75~
i) maximise area ABCO,
ii) minimise area CDEF,
iii) optimise permeate back-wash time CF,
iv) optimise gas back-wash tlme FG,
v) maximise recovered Elux GH, and,
vi) optimise permeate back-wash flux CD.
The gaseous back-wash can be implemented in a
number oE ways, and one such system is shown in Fig. 2. The
filter 10 has a draw-off line 11 through which permeate
normally flow~ to valve 12. Gas back-wash is introduced
through line 13 which includes a gas pressure control valve
14, a gas flow valve 15 and a gas on~off valve 16. The
filter 10 is connected to tank 17 through lines 18 and 19.
The inclusion of valve 15, which controls gas flow, gives
the line BD in Fig. 1 a sharp negative slope. The slower
the gas flow, the shallower the (negative) slope of that
line. If desired, the feed pressure may be dropped to zero
before back-wash, in which case the line BD will bend at C,
and ~D will ba vertical. This latter procedure is desirable
where the gas breakthrough pressures are high, so that a
lesser total pressure for the gas may be used since the
liquid feed pressure does not have to be overcome.
The gas back-wash time should be sufficient to
remove the fouling material from the membrane and from the
body of the filtar before re-application of normal process
conditions. In other words, the volume of expanded gas
should exceed the volume of the feed side of the hollow
fibre filter, or the ga9eous back-wash time should exceed
the residence time for feed flowing through the filter.
Gaseous back wash can be initiated automatically
by using a timer, or a flowswitch on the permeate line.
The following examples illustrate the application
o~ the invention and the methods used to give effect to the
invention of gaseous back-wash cleaning of ultrafilters.
.. - , . -
.. ~ ' .
..
,tJ.~3~ 4i~)
Exam~e 1
An ultrafilter of 0.16m2 area was made according
to Example 1 of our co-pending International patent
application by depositing a "cross-linked polyamide" within
the pores and as a network over the surface of a
polypropylene ultrafilter. Within the context of this
specification a "cross-linked polyamide" is one in which at
least one of the group consisting of acid halides and
primary and secondary amines is aromatic or substituted
aromatic and the acid halide is normally used in excess to
give chemical resistance and the average functionality is
above two so that there is considerable cross-linking.
In Example 1 of our International patent
application PCT/AU84/00179, terephthaloylchloride was used
as the acid chloride and bis ~3-aminopropyl) amine was used
because of its solubility in hexane to provide a deposition
formulation. The formulation was applied to the membrane
and polyamide deposition was complete within an hour.
In order to ensure that the material deposited
into the pores of the polypropylene a fine dilute emulsion
was formed of size about 1 micron and of such interfacial
tension with the continuous phase as to cause exclusion from
the smaller pores but to allow entry into the coarser
pores. The emulsion was formed as a hydrophobic mixture of:
Bis (3-aminopropyl) amine 3.93 grams
P-tertiaryoctylphenoxy-
polyethyleneglycolether 0.1 grams
Petroleum spirit (b.p. 60~80C) 950 millilitres
~bsolute ethanol 50 millilitres
Water was added drop-by drop until a distinct
opalescent turbidity indicated that droplets above the wave
length of visible light were present. These would, of
course, be above 1 micron in size. Care was taken to apply
the 1% weight per volume terephthaloylchloride solution in
petroleum spirit as soon as the amine solution had
evaporated to the desired ilm with droplets in the larger
~7S~(3
pores.
The trea-te,d membrane was washed in a 20% weight per
volume aqueous hydrochloric acid to dissolve any uncross-linked
material and to hydrolyse the excess terminal acid chloride to
carboxylic acid groups. A thorough water wash and drying at 60C
completed the treatment.
The filter was tested for permeability at 75kPa with
the following results:
Tap water flux was 372 l/m2hr.
Wheat starch factory flux
initial 220 l/m2hr.
after 30 min 162 1/m2hr.
after 45 min 124 1/m2hr.
after 16 hr 36 1/m2hr.
The seriously blocked ultrafil*er was then back-washed with
permeate at a pressure rising over 30 seconds -to 300kPa and
retested on the starch waste at 75kPa to give a flux of 151
1/m2hr. A test showed unchanged complete rejection of 0.1% BP
FEDAR0 M* soluble cutting oil. The Islight) stretch due
to the 200kPa lumen pressure was elastically recovered.
Although this experiment did show tha high pressure
permeate back-wash was enhanced by the stretching of the fibres
and that elastic recovery was possible, the loss of permeate was
a detracting feature.
Example 2
In another run on starch waste at lOOkPa, the above
cartridge ilter gave permeation rates after 24 hours of only 10
1/m2hr. Back-washing through the lumen with permeate at 150kPa
and 200kPa gave no increase in permeation ra~es~ However at
400kPa some slight swelling of the pores released particles which
resulted in an initial permeation rate of 34 1/m2hr on re-
testing. While the cleaning was still very incomplete there wasevidence for a steeply increasing benefit as internal pressures
rose to 400kPa. The reason for this was believed to be the
stretching of fibres.
* Trade Mark
.. . . :
''
'
5~i(3
~e 3
A larger composite cartridge (0.5m2) having hollow
fibres treated as set forth in Example 1 was fed with stiff
starch waste and showed that permeate back-wash did not
generate enough flow to clear the accumulation Erorn the
cartridge even at 700kPa. At this latter pressure, much
permeate was sacrificed in the 30 second flush back with
liquid. On switching to air back-wash, no air appeared in
five second trial pressures until 500kPa, when a sudden
cloud of fine bubbles rapidly turned to turbulent
decompression. The rapidly expanding gas increased five
fold in volume and swept out the shell side of the tubes
cleanly. When the shell side was kept running at 200kPa,
there was a need, as expected, to raise the pressure to
700kPa to achieve the same effect. The use of air back-wash
possessed the further advantage of minimising permeate loss.
Example 4
A composite cartridge of polypropylene made
hydrophilic with cross-linked polyamide gave 75~ rejection
of gelatin. It was used on a rapid fouling industrial waste
egg mucin. Initial rates at 1001cPa were 20 1/m2hr and these
fell off to 12 1/m2hr in 20 minutes. A back-blow with air
at 500kPa for five seconds returned rates to only 15
l/m2hr. It was noted that stringy mucin was removed
efficiently. It was found that a steadily rising back-wash
for five seconds with a small quantity of permeate left in
the line evidently cleared the microporous filling and that
the following air pulse at 500kPa peak blasted off the
stringy mucin.
In this case the small volume of permeate in the
permeate line (the minimum back-wash) steadily pressurised
above tha feed by 500kPa, followed by three seconds of
explosive gas expansion continually kept the rate at the
initial 20 1/m2hr, even when the feed was concentrated
threefold.
'~ '-
., .
s~
Exampl_ _
A 50 L sample of brine of density 1.36 was
obtained from a solar pond which had concentrated seawater
over a period of three years to a saturated condition. The
brine was contaminated by algae and other organic materials
which were required to be removed for crystallisation. The
sample was pumped into a cross-flow hollow polypropylene
fibre filtration cartridge. The inlet pressure was 270
kPa. The concentrate backpressure was 200 kPa and the
filtrate backpressure was 5 kPa. The initial flux was 106
L/hour from the cartridge. After 13 minutes this rate had
fallen to 64 L/hour and continued to decline to 50 L/hour
after a further 58 minutes. The unit was then back-washed
with air at 500 kPa and the flux immediately returned to ]06
L/hour. This cycle was repeated.
Example 6
A 50 L sample of hydrolized wheat starch was
filtered through a 50 micron screen and the fatty acids
decanted. The resulting liquor was highly turbid with a
very high suspended solids and dissolved solids loading.
This material was pumped into a hollow polypropylene fibre
cartridge at 200 kPa with 2 backpressure of 160 kPa and a
filtrate backpressure of 5 kPa. The initial flux of 58
L/hour declined after 24 minutes to 31 L/hour~ The
cartridge was then back-washed with 500 kPa air and the flux
immediately returned to 58 L/hour. Over a further period of
lS minutes the flux again declined to 31 L/hour. The
cartridge was again back-washed and the cycle repeated. At
all times the filtrate was a clear pale brown solution.
Example 7
The 30 L sample of black water soluble ink waste
from a packaging plant was pumped into a hollow
polypropylene fibre cartridge at 200 kPa with a backpressure
of 140 kPa and a filtrate backpressure of S kPa. The
initial flux on commencement was 82 L/hour from the
cartridge. Ater a period of 2S minutes this had declined
5~
-- 10 -
to 60 L~hour. The cartridge was then back-washed with 500
kPa air and the flux returned to 82. The cycle was
repeated. The Eiltrate at all times remained a clear pale
orange colour.
Example 8
A 50 L sample of polyvinyl acetate manufacturing
waste was obtained. This waste included PVA polymer as well
as other waste stream constituents. It had been treated by
the addition of ferric chloride and the addition of caustic
to adjust the pH to 10. The floc material was then pumped
into a hollow polypropylene fibre filtration cartridge at
185 kPa with a backpressure of 145 kPa. The initial
throughput rate was 53 L/hour. After 7 minutes it had
declined to 43 and after 20 minutes it had declined to 36
L/hour. The unit was then back-washed with 500 kPa air for
eight seconds and the flux immediately returned to 52
L/hour. The rate then continued to decline until back-wash
was repeated.
Example 9
A 25 L sample of unprocessed apple juice was
processed through a hollow polypropylene fibre
ultrafiltration unit~ The inlet pressure was 70 kPa, the
outlet pressure 65 kPa and the filtrate pressure 35 kPa. At
the commencement of filtration the flux was 420 L/hour and
~5 after a period of two hours this declined slcwly to 200
L/hour. On the application of a three second back-wash with
500 kPa air, the flux returned to 2~0 L/hour. After a
period of another hour the flux slowly declined to 200
L/hour and a three second back-wash again returned it to 240
L/hour. At all times the filtrate quality had a clarity
greater than that considered to be the best available in the
industry.
Example 10
A 25 L sample of water which had been used for
washing cut potatoes was collected for processing. The
initial input pressure was 200 kPa, backpressure on the
cartridge 150 kPa and filtrate pressure 20 kPa. When
filtration commenced the flux was 132 L/hour from the hollow
-. .. : : ' ' '
: - :' - "' "' " ' . .' ~ :'
' ~
~ ~75 ~ ~
polypropylene Eibre cartridge. AEter a period of 48
minutes, this flux declined to 90 L/hour. The cartridge was
back-washed with 500 kPa air for five seconds and the flux
returned to 120 L/hour. Over a period of another ten
minutes the Elux declined to 115 L/hour. The cartridge was
then back-washed for five seconds and the flux was restored
to 120 L/hour. The average potato starch content of the
feed material was 8.6% and the average potato starch
concentration in the filtrate was 0.13%.
Example 11
A sample of wheat starch hyrolysate was obtained
after it had been dosed with d~atomaceous earth at the rate
of 63 grams per 20 ~. This material was pumped into hollow
polypropylene fibre filtration cartridges at an inlet
pressure of 185 kPa, an outlet backpressure of 95 kPa and a
filtrate backpressure of 20 kPa. The initial throughput
rate of flux was 19.2 L/hour and after a period of 16
minutes this declined to 14.1 L/hour. The cartridges then
back-washed for 5 seconds with 475 kPa alr and the flux was
restored to 24.6 L/hour. The cartridge was then allowed to
continue filtration for a further 12 minutes and the flux
declined to 20.5 L/hour. Upon back-washing with air the
flux was restored to 25.8 L/hour.
Various modifications may be made in details of
the method of cleaning ultrafilters without departing from
the scope and ambit of the invention.
`
