- Email: [email protected]
Study of a hybrid process combining ozonation and membrane filtration — filtration of model solutions
DESALINATION Desalination 156 (2003) 257-265
ELSEVIER
www.elsevier.com/Iocate/desal
Study of a hybrid process combining ozonation and membrane filtration - filtration of model solutions B. Schlichter”, V. Mavrovb*, H. Chmiela~b “lnstitute,for
Environmentally Saarland
Compatible
University,
Tel. + 49 (681) 9345-340;
Process
Im Stadtwald
Technology
Ltd.,hDepartment
Geb. 47, D-66123
Fax + 49 (681) 9345380;
Saarbriicken,
email:
of Process
Technology
Germany
v,[email protected]
Received 7 February 2003; accepted 12 February 2003
Abstract
A hybrid process, consisting of ozonation and membrane filtration, was studied for the treatment of surface water or low-contaminated process wastewater. The main focus of this study centres on tests to determine the effect of ozone on membrane fouling during microfiltration and ultrafiltration. In this work, ceramic membranes with pore sizes in the range of 0.2 urn to a cut-off of 1 kD underwent lab-scale tests using model bentonite and humic acid solutions. Results showed that, by adding ozone during microfiltration and ultrafiltration ofthe humic acid solutions, membrane fouling for all membranes could be greatly reduced, thus obviating the need to backflush or clean the membranes. As for the bentonite solutions however, membrane fouling, which was caused by deposits of suspended inorganic substances, was not affected to any great extent by the addition of ozone. Microporous membranes with a pore size in the range of 1O-50 nm proved to be the best possible option for this hybrid process. Kqwords:
Ozonation; Micro/ultrafiltration; Bentonite; Humic acid; Membrane fouling
1. Introduction
For several years now, pressure-driven membrane processes such as microfiltration and ultrafiltration, operated singly or in combination with reverse osmosis and nanofiltration have been stateof-the-art technology in water and wastewater *Corresponding author. Presented
ot the European
European
Desalination
00 I I-91 64103/$-
See
Conference
Society
on Desalination
International
treatment. These processes are predominantly applied in drinking water and process water production [I] as well as in the treatment of industrial wastewater [2] (Fig. I). Compared to conventional separation processes, the most significant advantage of these processes, besides the reliable separation of suspended and colloidal organic and inorganic
and the Environment:
Fresh Water for All,
Water Association.
front matter 0 2003 Elsevier Science B.V. All rights reserved
PII: SO01 1-9 164(03)00348-5
Malta.
4-8 May 2003.
B. Schlichtes et al.
258
i
Desalination 156 (2003) 257 -~265
+ Concentrated wastewater
+ Concentrated wastewater
Dismfection
Surface water
1
-
b
Drmking water or industrial process water
+ Concentrated wastewater
Fig. I. Overview of typical process variants for water treatment by pressure-driven membrane processes.
substances, is the retention of pathogenic bacteria and every type of micro-organism to a great extent. However, one of the major disadvantages of these membrane processes is fouling or biofouling of the membrane induced by deposits of inorganic, organic and microbiological substances on both the membrane surface and inside the membrane pores. Extensive membrane fouling leads to a pronounced decrease in permeate flux and can threaten the economic efficiency of the membrane plant. Different measures are implemented in order to limit membrane fouling of microfiltration and ultrafiltration membranes in practical applications. The most common are: Periodical membrane backflushing Regular chemical membrane cleaning Water pretreatment ??
??
??
Nevertheless, there are many disadvantages linked to these measures. The periodical backflushing of the membranes causes the total yield of the process to be reduced generally to between 70 and 90% [3], resulting in a wastewater quantity of 0.1 I to 0.43 m’ for disposal for each cubic meter of fresh water produced. Since this wastewater is in concentrated form, which often contains the added cleaning and backflushing chemicals besides the substances to be separated, the dis-
posal of this wastewater stream is not always without problems and frequently incurs considerable disposal costs. Pretreatment of the water to be filtered can also be expensive because additional investment and operating costs can be involved depending on the treatment process in operation. Furthermore, if fouling control is to be effective, chlorine cleaning agents are generally used which results not only in high chemical consumption but also frequently in the formation of unwanted chlorine byproducts [4]. When applying microfiltration or ultrafiltration (MF/UF), as a pretreatment process (Fig. I), the permeate generated in this stage has to fulftl specific quality requirements so as the downstream nanofiltration or reverse osmosis (NF/RO) process can operate successfully. Particularly microorganisms such as bacteria and viruses are separated, to a great extent, by microfiltration or ultrafiltration but are not completely eliminated. This results in an increase of microorganism concentration in the feed water of the downstream NF/RO stage which can reduce the efficiency ofthis second filtration stage considerably due to biofouling. This study focuses on the development and testing of a new hybrid process for water treatment combining ozonation and microfiltration/ultrafiltration (Fig. 2), having as its objective, the
259
B. Schlichter et al. 1 Desalination 156 (2003) 257-265
Treated water
Surface water or industrial wastewater
Minimised wastewater quantity
Fig. 2. Schematic representationofthe new hybrid processfor water treatment combining ozonation and membrane filtration.
minimisation of membrane fouling and its consequences, in order to increase the performance and economic efficiency of this membrane process. III this work, the filtration of model solutions
containing suspended inorganic substances (bentonite solutions) as well as solutions containing dissolved and partially colloidally dissolved organic components (humic acid solutions) is investigated in order to determine the effect of ozone on the fouling behaviour of membranes.
2. Experimental 2.1. Membranes Commercial, ceramic multi-channel microfiltration and ultrafiltration tubular membranes with an outer diameter of 25.5 mm (kO.5) respectively and lengths of 200, 500 and 1000 mm (Table 1) were used in membrane filtration operations. The filtration direction was from the inside to the outside of the ceramic membrane element. Table
2.2. Experimental
set-up
The test set-up, as shown in Fig. 3, was used to perform the filtration experiments. All the filtration tests were conducted in the crossflow mode, the gaseous ozone being added via an injector (Venturi tube) into the feed flow prior to membrane filtration. To ensure an optimum mass transfer of the ozone, the water was pumped through a static mixer directly after the ozone was added. Butterfly valves were used to control bypass, feed, retentate and permeate volume flows in order to set up the required filtration conditions, in particular overflow velocity and transmembrane pressure. Bentonite (Fluka) with an average particle size of 3 pm in aqueous solutions, in a concentration of 0.1 g/l and with a turbidity of 8.0 NTU as well as humic acid (Aldrich) in a concentration of 50 mg/l were used in the model solutions. Model humic acid solutions were characterised as follows: TOC 19.5 mg/l, COD (potassium permanganate) 25.8 mgO,/l, turbidity 12.1 NTU and
I
Propertiesof the ceramic microfiltration and ultrafiltration membranes under study [6] Synonym
MF 1 MF2 MF3 UF I UF2 UF3 -.___
Geometry
Pore size/ cut-off
Membrane area, m*
Pure water permeability, I/(m’.h.bar)
19 channel 19 channel 19 channel 19 channel 23 channel 23 channel
0.20 pm 0.10 pm 0.05 pm 20 kD (-1 Onm) 15 kD (-6nm) 1 kD (-4nm)
0.0335 0.0335 0.0960 0.2000 0.283 1 0.283 1
150&1600 600-650 350-370 190-210 95-100 25-28
260
R Schlichter
et al.
I Desalination ~
156 (2003)
257.-265
Permeate reclrculatlon
Model solutions / water to be treated
“f’
Permeate
Tzd
water
Membrane module V
v Reactor vessel
Ozone gas Static mixer Injector
Fig. 3. Simplified flow diagram of the experimental set-up.
pH-7. Model solutions were prepared using fully desalinated water.
3. Results and discussion 3. I. Menlbrane,filtration
of bentonite solutions
To study the effect of ozone on membrane fouling caused by particulate inorganic substances, the steady-state permeate fluxes generated by the filtration of bentonite solutions without ozone, as well as for bentonite solutions with ozone concentrations of I mg/l at various transmembrane pressures ranging between 0 and 1.O bar were determined. Steady-state conditions became established after five hours of filtration. The results are illustrated in Fig. 4 and compared to the pure water fluxes of each membrane respectively. For all filtration tests, the flow conditions were comparable, flow velocity being 0.7 m/s and there was complete membrane retention of the bentonite particles (turbidity in the permeate was below 0.1 NTU). During the filtration of the bentonite solution containing ozone, the presence of ozone had no distinct influence on permeation behaviour and the resulting steady-state permeate fluxes. In this
case and in the filtration of bentonite solution without ozone, it was not possible, for all membranes except UF 3, to increase the permeate flux by increasing the transmembrane pressure above approx. 0.5 bar, because the critical flux had been reached. It can be assumed that fouling had occurred due to deposits of bentonite particles on the membrane surfaces. In the tests conducted by Madaeni [5] involving latex particles under similar filtration conditions, it was observed that, once the critical flux had been reached, a noticeable coating layer had formed on the membrane surface whereas below the critical flux, only individual pores became blocked. These facts could be verified by electron micrographs of the membranes used in this study 163. Concerning the filtration experiments with bentonite solutions performed in this study. it can be concluded that the presence of ozone had no major influence on pore blocking nor on coating layer formation caused by bentonite particles. 3.2. Membrane, filtration of humic acid solutions III order to compare the fouling behaviour of the microfiltration and ultrafiltration membranes for the model humic acid solutions with and
B. Schlichter
et al. / Desalination
156 (2003)
261
257-265
1
-mf*
-n-
pure water
pure water + ozone bentonite 0.1 g/l + ozone
bentonite 0.1 g/l 250
1200
200.
900
150.
UF
I
I(20 kD)
600 300 2
0 .~....._."_~.~..,___~.:.....__~...~_..~,_._.....c_................P.~
25 20 15 IO 5 0 0.0
transmembrane
0,2
pressure
0.4
0,6
0,8
I,0
[bar]
Fig. 4. Steady-state permeate fluxes as a function of transmembrane pressure for the microfiltration and ultrafiltration of pure water and bentonite solutions with and without the addition of ozone.
without the addition of ozone (ozone dose of -2.6 mg O,/mg TOC), tests were conducted to determine the steady-state permeate fluxes as a function of transmembrane pressure (Fig. 5). The filtration conditions were the same as those during the filtration tests with bentonite. The trends of these curves were compared to the respective linear pure water fluxes. In contrast to the filtration tests involving humic acid solutions without ozone, higher steady-state permeate
fluxes were determined for all microtiltration and ultrafiltration membranes during the filtration of humic acid solutions to which ozone had been added. The only exception was the ultrafiltration membrane UF 3, which, due to its low cut-off of 1 kD, showed no distinct membrane fouling for the humic acids without ozone in the pressure range in question. Therefore there was no further significant increase in permeate flux in this case when ozone was added.
262.
B. Schlichter
et al.
’Desalination
156 (2003)
257-265
“__,____________II~“~~.~“ ...-- _- _____.^.^_. i I
I
.____^ -•--
-•-
pure water
,%: humic acid
I
pure water + ozone
50 mgil
humic acid 50 mgll + ozone
^^ __^._ “_x.^” _.... _^_._” ,_,,_“.___
1500 -. -- --
...” ..__._.. -... .__..” ..“^
250
x.0 ,200~
MF 1 (0.20 IJm).;-i1
200
900
150
GOO-
.g”
100
300. F= C ?$
=
5
O-<
50 .*
/f-
L
L)
*
dtl 0
.- -. .-.-.-. I ----.....---..
76O”, _-._..
600. MF2 (0.10
100
pm)
60.
E
/
/
UF 2 (15 kD)
60.
al
jjE
jR 300 150
dP
,,.r"<
':
*
a'
,LI
o3025-
UF 3 (1 kD)
20.
0,o
transmembrane
0.2
pressure
0.4
0.6
0.8
1.0
[bar]
Fig. 5. Steady-state permeate fluxes as a function of transmembrane pressure for the microfiltration and ultrafiltration of pure water and humic acid solutions with and without the addition of ozone.
As Fig. 5 shows, no further increase in permeate flux occurred during the microfiltration tests with humic acid solutions without ozone involving membranes with a cut-off of between 0.05 and 0.20 pm at a transmembrane pressure of above 0.4 bar. The cause of this was deposits of hmnic acid on the microfiltration membrane under study which finally resulted in constantly low permeate fluxes independent of pressure level. In contrast, permeate fluxes, which in-
creased I inear to transmembrane pressure, were obtained (with the exception of MF l), in the microfiltration tests using humic acid solutions containing ozone. However the increase was not as steep as for pure water permeabilities which can be attributed not only to partial pore blocking caused by traces of particles present in the humic acid solution but also to polarisation phenomena [7]. In the case of the membrane MF 1, the addition of ozone caused comparatively higher steady-
R. Schlichter
et al. i Desalination
90
263
257-265
at 1 kD (4 nm) -the permeate fluxes for humic acid filtration with the addition of ozone rise from 20% to above 90% of the pure water flux. To judge the effectiveness of ozone dosage during membrane filtration, the proportion of the permeate flux, which could be ascribed to the addition of ozone, was illustrated (Fig. 6). These proportions were calculated from the differences of the permeate fluxes for the filtration of humic acids with and without ozone. The bar chart highlights that the most effective results were achieved for the two microporous membranes MF 3 and UF 1 with pore sizes ranging from 10 to 50 nm. The membranes reached approximately 70-80% of the pure water flux, in which case roughly 50% can be attributed to the addition of ozone. Figs. 7 and 8 show the filtration time trends of the normalised permeabilities (membrane permeabilty to pure water permeability) during humic acid filtration with and without ozone followed by the filtration of fully desalinated water (pure water) with and without ozone for the membranes MF 2 and UF 2. Considerable reductions in permeability can be seen for both microfiltration and ultrafiltration membranes during the initial 170 min of humic acid filtration (Fig. 7). After 170 min of filtration
state permeate fluxes compared with the filtration of humic acid solution without ozone but at a transmembrane pressure of above 0.5 bar in both cases (with and without ozone), no further increase in permeate flux occurred because the critical flux had been reached. The results of the tests for ultrafiltration were similar to those for microfiltration. It could also be determined that as the cut-off of the ultrafiltration membrane decreased, fouling occurred to a lesser extent. The reductions in permeate flux observed for the membrane UF 3 with a cut-off of I kD were so minimal that this could mainly be attributed to concentration polarisation. The most important result of these tests is that the addition of ozone during the filtration of the humic acids more or less causes a significant increase in permeate fluxes. The steady-state permeate fluxes were plotted (Fig. 6) as a function of membrane pore size/cut-off in order to determine which membranes showed the greatest increase in flux once ozone had been added. The filtration tests were conducted for all the membranes under the same conditions, i.e. at a transmembrane pressure of I .O bar and a specific ozone dosage of 2.6 mg/mg TOC. It can be seen that (Fig. 6) as pore size decreases -from microfiltration at 0.2 pm to ultrafiltration 100
156 (2003)
/
’ / ??Proportion of permeate flux due to ozone addition
0.2 pm W 1)
0 1 pm WF 2)
0.05 pm W 3)
20 kD KJF 1)
15kD (‘JF 2)
1 kD W 3)
Membrane pore size/cut-off Fig. 6. The influence of ozone addition on the permeate fluxes during humic acid filtration as a function of membrane pore size/cut-off.
264
B Schlichter
et al. I Desalination
IS6 (2003) 257-265
humic acid s&.ttion -O-filtration
t
without
w*ater
ozone
-O--filtration
with ozone
ozonation
04-
::$ 0
y=;_(15,y)
40
,
80
I
140
8
I
160
‘backflushing r
180
I
8
200
1
T
220
filtration time [min] Fig. 7. Tests on the cleaning of membranes MF 2 and UF 2 after humic acid filtration by backflushing and addition of ozone.
-O-filtration
x z _Q
with ozone
-o-
filtration
without
ozone
1.0
zE
08
& a
0.6
-0 % I=
0.4
E
0.2
c
00 0
40
60
140
160
180
200
220
filtration time [min] Fig. 8. Tests on membrane backflushing following the filtration of humic acid solution with the addition of ozone
was replaced by fully desalinated water and filtration continued under the same conditions. After 180 min, both membranes were backflushed with fillly desalinated water for two minutes. Finally ozone was added to the pure water to be filtered after 200 min. The ozone concentration in the feed water was roughly l-2 mg/l. Judging from the resulting permeability trends,
time, the humic acid solution
the permeability ofthe microfiltration membrane (MF 2) could be restored to 60-70% of its original pure water permeability by membrane backflushing with pure water. whereas the permeability of the ultrafiltration membrane (UF 2) remained unchanged after backflushing. This indicated that, in filtration without ozone, MF membrane fouling is due to adsorption, mechanical
B. Schlichter
et al. / Desalination
pore blocking and coating layer formation whereas in the case of the UF membrane, fouling is caused mainly by adsorption. Fig. 8 shows the permeability trend of a microfiltration membrane (MF 2) during humic acid filtration involving ozone. After 170 min, the humic acid was again replaced by fully desalinated water and filtration continued without further addition of ozone for ten minutes. The membrane was then backflushed with fully desalinated water for two minutes and the filtration operation continued using fully desalinated water. This showed that it was possible to re-establish the pure water permeability of the membrane. It can be concluded that, during humic acid filtration with ozone, fouling was mainly caused by mechanical pore blocking and coating layer formation but not by adsorption.
4. Conclusions Adding ozone during the microfiltration and ultrafiltration of pure water had no influence on filtration behaviour i.e. neither an increase nor decrease in permeate flux was observed. The filtration experiments with bentonite solutions showed that membrane fouling is governed mostly by coating layer formation and remains unaffected when ozone is continuously added i.e. the same critical flux was determined for all the membranes respectively for the filtration of bentonite solutions containing or not containing ozone. In contrast, the filtration of humic acid solutions produced clearly different results; the presence of ozone caused a distinct increase in critical flux under otherwise similar filtration conditions i.e. at the
156
(2003) 257-265
265
same temperature, ozone concentration and flow velocity. During the microfiltration and ultrafiltration of humic acid solutions, the presence of ozone caused a drastic reduction in adsorptioninduced membrane fouling, mostly resulting from the adsorption of organic substances. Particularly for membranes with pore sizes of between 10 and 50 nm, the continuous addition of ozone proved to be the most effective fouling control measure.
References M. Clever, F. Jordt, R. Knauf, N. Rabiger, M. Rtidebusch and R. Hilker-Scheibel, Process water production from river water by ultrafiltration and reverse osmosis, Desalination, 13 1 (2000) 325-336. H. 121 Chmiel, V. Mavrov and E. Belihes, Reuse of vapour condensate from milk processing using nanofiltration, Filtr. Separ., 4 (2000) 24-27. [31 M. Bodzek and K. Konieczny, Comparision of various membrane types and module configurations in the treatment of natural water by means of low pressure membrane methods, Separ. Purif. Technol., 14 (1998) 69-78. 141 W. Coleman, J. Munch and P. Ringhand, Ozonation/ post-chlorination of humic acid: a model for predicting drinking water disinfection by-products, Ozone Sci. Eng., 14 (1992) 51-69. S.S. Madaeni, Investigation of the mechanism of [51 critical flux in membrane filtration using electron microscopy, J. Porous Materials, 4 (1997) 239-244. [61 B. Schlichter, Entwicklung und Untersuchung eines neuartigen Wasseraufbereitungsverfahrens bestehend aus Wasserozonung und Mikro-/ Ultrafiltration, Mensch & Buch Verlag, Berlin, 2003 [in press]. H. De Balmann and V, Sanchez, Coupling between [71 concentration polarisation and fouling in ultrafiltration, Intnl. J. Chem. Eng., 4(34) (1992) 665673. [II
ELSEVIER
www.elsevier.com/Iocate/desal
Study of a hybrid process combining ozonation and membrane filtration - filtration of model solutions B. Schlichter”, V. Mavrovb*, H. Chmiela~b “lnstitute,for
Environmentally Saarland
Compatible
University,
Tel. + 49 (681) 9345-340;
Process
Im Stadtwald
Technology
Ltd.,hDepartment
Geb. 47, D-66123
Fax + 49 (681) 9345380;
Saarbriicken,
email:
of Process
Technology
Germany
v,[email protected]
Received 7 February 2003; accepted 12 February 2003
Abstract
A hybrid process, consisting of ozonation and membrane filtration, was studied for the treatment of surface water or low-contaminated process wastewater. The main focus of this study centres on tests to determine the effect of ozone on membrane fouling during microfiltration and ultrafiltration. In this work, ceramic membranes with pore sizes in the range of 0.2 urn to a cut-off of 1 kD underwent lab-scale tests using model bentonite and humic acid solutions. Results showed that, by adding ozone during microfiltration and ultrafiltration ofthe humic acid solutions, membrane fouling for all membranes could be greatly reduced, thus obviating the need to backflush or clean the membranes. As for the bentonite solutions however, membrane fouling, which was caused by deposits of suspended inorganic substances, was not affected to any great extent by the addition of ozone. Microporous membranes with a pore size in the range of 1O-50 nm proved to be the best possible option for this hybrid process. Kqwords:
Ozonation; Micro/ultrafiltration; Bentonite; Humic acid; Membrane fouling
1. Introduction
For several years now, pressure-driven membrane processes such as microfiltration and ultrafiltration, operated singly or in combination with reverse osmosis and nanofiltration have been stateof-the-art technology in water and wastewater *Corresponding author. Presented
ot the European
European
Desalination
00 I I-91 64103/$-
See
Conference
Society
on Desalination
International
treatment. These processes are predominantly applied in drinking water and process water production [I] as well as in the treatment of industrial wastewater [2] (Fig. I). Compared to conventional separation processes, the most significant advantage of these processes, besides the reliable separation of suspended and colloidal organic and inorganic
and the Environment:
Fresh Water for All,
Water Association.
front matter 0 2003 Elsevier Science B.V. All rights reserved
PII: SO01 1-9 164(03)00348-5
Malta.
4-8 May 2003.
B. Schlichtes et al.
258
i
Desalination 156 (2003) 257 -~265
+ Concentrated wastewater
+ Concentrated wastewater
Dismfection
Surface water
1
-
b
Drmking water or industrial process water
+ Concentrated wastewater
Fig. I. Overview of typical process variants for water treatment by pressure-driven membrane processes.
substances, is the retention of pathogenic bacteria and every type of micro-organism to a great extent. However, one of the major disadvantages of these membrane processes is fouling or biofouling of the membrane induced by deposits of inorganic, organic and microbiological substances on both the membrane surface and inside the membrane pores. Extensive membrane fouling leads to a pronounced decrease in permeate flux and can threaten the economic efficiency of the membrane plant. Different measures are implemented in order to limit membrane fouling of microfiltration and ultrafiltration membranes in practical applications. The most common are: Periodical membrane backflushing Regular chemical membrane cleaning Water pretreatment ??
??
??
Nevertheless, there are many disadvantages linked to these measures. The periodical backflushing of the membranes causes the total yield of the process to be reduced generally to between 70 and 90% [3], resulting in a wastewater quantity of 0.1 I to 0.43 m’ for disposal for each cubic meter of fresh water produced. Since this wastewater is in concentrated form, which often contains the added cleaning and backflushing chemicals besides the substances to be separated, the dis-
posal of this wastewater stream is not always without problems and frequently incurs considerable disposal costs. Pretreatment of the water to be filtered can also be expensive because additional investment and operating costs can be involved depending on the treatment process in operation. Furthermore, if fouling control is to be effective, chlorine cleaning agents are generally used which results not only in high chemical consumption but also frequently in the formation of unwanted chlorine byproducts [4]. When applying microfiltration or ultrafiltration (MF/UF), as a pretreatment process (Fig. I), the permeate generated in this stage has to fulftl specific quality requirements so as the downstream nanofiltration or reverse osmosis (NF/RO) process can operate successfully. Particularly microorganisms such as bacteria and viruses are separated, to a great extent, by microfiltration or ultrafiltration but are not completely eliminated. This results in an increase of microorganism concentration in the feed water of the downstream NF/RO stage which can reduce the efficiency ofthis second filtration stage considerably due to biofouling. This study focuses on the development and testing of a new hybrid process for water treatment combining ozonation and microfiltration/ultrafiltration (Fig. 2), having as its objective, the
259
B. Schlichter et al. 1 Desalination 156 (2003) 257-265
Treated water
Surface water or industrial wastewater
Minimised wastewater quantity
Fig. 2. Schematic representationofthe new hybrid processfor water treatment combining ozonation and membrane filtration.
minimisation of membrane fouling and its consequences, in order to increase the performance and economic efficiency of this membrane process. III this work, the filtration of model solutions
containing suspended inorganic substances (bentonite solutions) as well as solutions containing dissolved and partially colloidally dissolved organic components (humic acid solutions) is investigated in order to determine the effect of ozone on the fouling behaviour of membranes.
2. Experimental 2.1. Membranes Commercial, ceramic multi-channel microfiltration and ultrafiltration tubular membranes with an outer diameter of 25.5 mm (kO.5) respectively and lengths of 200, 500 and 1000 mm (Table 1) were used in membrane filtration operations. The filtration direction was from the inside to the outside of the ceramic membrane element. Table
2.2. Experimental
set-up
The test set-up, as shown in Fig. 3, was used to perform the filtration experiments. All the filtration tests were conducted in the crossflow mode, the gaseous ozone being added via an injector (Venturi tube) into the feed flow prior to membrane filtration. To ensure an optimum mass transfer of the ozone, the water was pumped through a static mixer directly after the ozone was added. Butterfly valves were used to control bypass, feed, retentate and permeate volume flows in order to set up the required filtration conditions, in particular overflow velocity and transmembrane pressure. Bentonite (Fluka) with an average particle size of 3 pm in aqueous solutions, in a concentration of 0.1 g/l and with a turbidity of 8.0 NTU as well as humic acid (Aldrich) in a concentration of 50 mg/l were used in the model solutions. Model humic acid solutions were characterised as follows: TOC 19.5 mg/l, COD (potassium permanganate) 25.8 mgO,/l, turbidity 12.1 NTU and
I
Propertiesof the ceramic microfiltration and ultrafiltration membranes under study [6] Synonym
MF 1 MF2 MF3 UF I UF2 UF3 -.___
Geometry
Pore size/ cut-off
Membrane area, m*
Pure water permeability, I/(m’.h.bar)
19 channel 19 channel 19 channel 19 channel 23 channel 23 channel
0.20 pm 0.10 pm 0.05 pm 20 kD (-1 Onm) 15 kD (-6nm) 1 kD (-4nm)
0.0335 0.0335 0.0960 0.2000 0.283 1 0.283 1
150&1600 600-650 350-370 190-210 95-100 25-28
260
R Schlichter
et al.
I Desalination ~
156 (2003)
257.-265
Permeate reclrculatlon
Model solutions / water to be treated
“f’
Permeate
Tzd
water
Membrane module V
v Reactor vessel
Ozone gas Static mixer Injector
Fig. 3. Simplified flow diagram of the experimental set-up.
pH-7. Model solutions were prepared using fully desalinated water.
3. Results and discussion 3. I. Menlbrane,filtration
of bentonite solutions
To study the effect of ozone on membrane fouling caused by particulate inorganic substances, the steady-state permeate fluxes generated by the filtration of bentonite solutions without ozone, as well as for bentonite solutions with ozone concentrations of I mg/l at various transmembrane pressures ranging between 0 and 1.O bar were determined. Steady-state conditions became established after five hours of filtration. The results are illustrated in Fig. 4 and compared to the pure water fluxes of each membrane respectively. For all filtration tests, the flow conditions were comparable, flow velocity being 0.7 m/s and there was complete membrane retention of the bentonite particles (turbidity in the permeate was below 0.1 NTU). During the filtration of the bentonite solution containing ozone, the presence of ozone had no distinct influence on permeation behaviour and the resulting steady-state permeate fluxes. In this
case and in the filtration of bentonite solution without ozone, it was not possible, for all membranes except UF 3, to increase the permeate flux by increasing the transmembrane pressure above approx. 0.5 bar, because the critical flux had been reached. It can be assumed that fouling had occurred due to deposits of bentonite particles on the membrane surfaces. In the tests conducted by Madaeni [5] involving latex particles under similar filtration conditions, it was observed that, once the critical flux had been reached, a noticeable coating layer had formed on the membrane surface whereas below the critical flux, only individual pores became blocked. These facts could be verified by electron micrographs of the membranes used in this study 163. Concerning the filtration experiments with bentonite solutions performed in this study. it can be concluded that the presence of ozone had no major influence on pore blocking nor on coating layer formation caused by bentonite particles. 3.2. Membrane, filtration of humic acid solutions III order to compare the fouling behaviour of the microfiltration and ultrafiltration membranes for the model humic acid solutions with and
B. Schlichter
et al. / Desalination
156 (2003)
261
257-265
1
-mf*
-n-
pure water
pure water + ozone bentonite 0.1 g/l + ozone
bentonite 0.1 g/l 250
1200
200.
900
150.
UF
I
I(20 kD)
600 300 2
0 .~....._."_~.~..,___~.:.....__~...~_..~,_._.....c_................P.~
25 20 15 IO 5 0 0.0
transmembrane
0,2
pressure
0.4
0,6
0,8
I,0
[bar]
Fig. 4. Steady-state permeate fluxes as a function of transmembrane pressure for the microfiltration and ultrafiltration of pure water and bentonite solutions with and without the addition of ozone.
without the addition of ozone (ozone dose of -2.6 mg O,/mg TOC), tests were conducted to determine the steady-state permeate fluxes as a function of transmembrane pressure (Fig. 5). The filtration conditions were the same as those during the filtration tests with bentonite. The trends of these curves were compared to the respective linear pure water fluxes. In contrast to the filtration tests involving humic acid solutions without ozone, higher steady-state permeate
fluxes were determined for all microtiltration and ultrafiltration membranes during the filtration of humic acid solutions to which ozone had been added. The only exception was the ultrafiltration membrane UF 3, which, due to its low cut-off of 1 kD, showed no distinct membrane fouling for the humic acids without ozone in the pressure range in question. Therefore there was no further significant increase in permeate flux in this case when ozone was added.
262.
B. Schlichter
et al.
’Desalination
156 (2003)
257-265
“__,____________II~“~~.~“ ...-- _- _____.^.^_. i I
I
.____^ -•--
-•-
pure water
,%: humic acid
I
pure water + ozone
50 mgil
humic acid 50 mgll + ozone
^^ __^._ “_x.^” _.... _^_._” ,_,,_“.___
1500 -. -- --
...” ..__._.. -... .__..” ..“^
250
x.0 ,200~
MF 1 (0.20 IJm).;-i1
200
900
150
GOO-
.g”
100
300. F= C ?$
=
5
O-<
50 .*
/f-
L
L)
*
dtl 0
.- -. .-.-.-. I ----.....---..
76O”, _-._..
600. MF2 (0.10
100
pm)
60.
E
/
/
UF 2 (15 kD)
60.
al
jjE
jR 300 150
dP
,,.r"<
':
*
a'
,LI
o3025-
UF 3 (1 kD)
20.
0,o
transmembrane
0.2
pressure
0.4
0.6
0.8
1.0
[bar]
Fig. 5. Steady-state permeate fluxes as a function of transmembrane pressure for the microfiltration and ultrafiltration of pure water and humic acid solutions with and without the addition of ozone.
As Fig. 5 shows, no further increase in permeate flux occurred during the microfiltration tests with humic acid solutions without ozone involving membranes with a cut-off of between 0.05 and 0.20 pm at a transmembrane pressure of above 0.4 bar. The cause of this was deposits of hmnic acid on the microfiltration membrane under study which finally resulted in constantly low permeate fluxes independent of pressure level. In contrast, permeate fluxes, which in-
creased I inear to transmembrane pressure, were obtained (with the exception of MF l), in the microfiltration tests using humic acid solutions containing ozone. However the increase was not as steep as for pure water permeabilities which can be attributed not only to partial pore blocking caused by traces of particles present in the humic acid solution but also to polarisation phenomena [7]. In the case of the membrane MF 1, the addition of ozone caused comparatively higher steady-
R. Schlichter
et al. i Desalination
90
263
257-265
at 1 kD (4 nm) -the permeate fluxes for humic acid filtration with the addition of ozone rise from 20% to above 90% of the pure water flux. To judge the effectiveness of ozone dosage during membrane filtration, the proportion of the permeate flux, which could be ascribed to the addition of ozone, was illustrated (Fig. 6). These proportions were calculated from the differences of the permeate fluxes for the filtration of humic acids with and without ozone. The bar chart highlights that the most effective results were achieved for the two microporous membranes MF 3 and UF 1 with pore sizes ranging from 10 to 50 nm. The membranes reached approximately 70-80% of the pure water flux, in which case roughly 50% can be attributed to the addition of ozone. Figs. 7 and 8 show the filtration time trends of the normalised permeabilities (membrane permeabilty to pure water permeability) during humic acid filtration with and without ozone followed by the filtration of fully desalinated water (pure water) with and without ozone for the membranes MF 2 and UF 2. Considerable reductions in permeability can be seen for both microfiltration and ultrafiltration membranes during the initial 170 min of humic acid filtration (Fig. 7). After 170 min of filtration
state permeate fluxes compared with the filtration of humic acid solution without ozone but at a transmembrane pressure of above 0.5 bar in both cases (with and without ozone), no further increase in permeate flux occurred because the critical flux had been reached. The results of the tests for ultrafiltration were similar to those for microfiltration. It could also be determined that as the cut-off of the ultrafiltration membrane decreased, fouling occurred to a lesser extent. The reductions in permeate flux observed for the membrane UF 3 with a cut-off of I kD were so minimal that this could mainly be attributed to concentration polarisation. The most important result of these tests is that the addition of ozone during the filtration of the humic acids more or less causes a significant increase in permeate fluxes. The steady-state permeate fluxes were plotted (Fig. 6) as a function of membrane pore size/cut-off in order to determine which membranes showed the greatest increase in flux once ozone had been added. The filtration tests were conducted for all the membranes under the same conditions, i.e. at a transmembrane pressure of I .O bar and a specific ozone dosage of 2.6 mg/mg TOC. It can be seen that (Fig. 6) as pore size decreases -from microfiltration at 0.2 pm to ultrafiltration 100
156 (2003)
/
’ / ??Proportion of permeate flux due to ozone addition
0.2 pm W 1)
0 1 pm WF 2)
0.05 pm W 3)
20 kD KJF 1)
15kD (‘JF 2)
1 kD W 3)
Membrane pore size/cut-off Fig. 6. The influence of ozone addition on the permeate fluxes during humic acid filtration as a function of membrane pore size/cut-off.
264
B Schlichter
et al. I Desalination
IS6 (2003) 257-265
humic acid s&.ttion -O-filtration
t
without
w*ater
ozone
-O--filtration
with ozone
ozonation
04-
::$ 0
y=;_(15,y)
40
,
80
I
140
8
I
160
‘backflushing r
180
I
8
200
1
T
220
filtration time [min] Fig. 7. Tests on the cleaning of membranes MF 2 and UF 2 after humic acid filtration by backflushing and addition of ozone.
-O-filtration
x z _Q
with ozone
-o-
filtration
without
ozone
1.0
zE
08
& a
0.6
-0 % I=
0.4
E
0.2
c
00 0
40
60
140
160
180
200
220
filtration time [min] Fig. 8. Tests on membrane backflushing following the filtration of humic acid solution with the addition of ozone
was replaced by fully desalinated water and filtration continued under the same conditions. After 180 min, both membranes were backflushed with fillly desalinated water for two minutes. Finally ozone was added to the pure water to be filtered after 200 min. The ozone concentration in the feed water was roughly l-2 mg/l. Judging from the resulting permeability trends,
time, the humic acid solution
the permeability ofthe microfiltration membrane (MF 2) could be restored to 60-70% of its original pure water permeability by membrane backflushing with pure water. whereas the permeability of the ultrafiltration membrane (UF 2) remained unchanged after backflushing. This indicated that, in filtration without ozone, MF membrane fouling is due to adsorption, mechanical
B. Schlichter
et al. / Desalination
pore blocking and coating layer formation whereas in the case of the UF membrane, fouling is caused mainly by adsorption. Fig. 8 shows the permeability trend of a microfiltration membrane (MF 2) during humic acid filtration involving ozone. After 170 min, the humic acid was again replaced by fully desalinated water and filtration continued without further addition of ozone for ten minutes. The membrane was then backflushed with fully desalinated water for two minutes and the filtration operation continued using fully desalinated water. This showed that it was possible to re-establish the pure water permeability of the membrane. It can be concluded that, during humic acid filtration with ozone, fouling was mainly caused by mechanical pore blocking and coating layer formation but not by adsorption.
4. Conclusions Adding ozone during the microfiltration and ultrafiltration of pure water had no influence on filtration behaviour i.e. neither an increase nor decrease in permeate flux was observed. The filtration experiments with bentonite solutions showed that membrane fouling is governed mostly by coating layer formation and remains unaffected when ozone is continuously added i.e. the same critical flux was determined for all the membranes respectively for the filtration of bentonite solutions containing or not containing ozone. In contrast, the filtration of humic acid solutions produced clearly different results; the presence of ozone caused a distinct increase in critical flux under otherwise similar filtration conditions i.e. at the
156
(2003) 257-265
265
same temperature, ozone concentration and flow velocity. During the microfiltration and ultrafiltration of humic acid solutions, the presence of ozone caused a drastic reduction in adsorptioninduced membrane fouling, mostly resulting from the adsorption of organic substances. Particularly for membranes with pore sizes of between 10 and 50 nm, the continuous addition of ozone proved to be the most effective fouling control measure.
References M. Clever, F. Jordt, R. Knauf, N. Rabiger, M. Rtidebusch and R. Hilker-Scheibel, Process water production from river water by ultrafiltration and reverse osmosis, Desalination, 13 1 (2000) 325-336. H. 121 Chmiel, V. Mavrov and E. Belihes, Reuse of vapour condensate from milk processing using nanofiltration, Filtr. Separ., 4 (2000) 24-27. [31 M. Bodzek and K. Konieczny, Comparision of various membrane types and module configurations in the treatment of natural water by means of low pressure membrane methods, Separ. Purif. Technol., 14 (1998) 69-78. 141 W. Coleman, J. Munch and P. Ringhand, Ozonation/ post-chlorination of humic acid: a model for predicting drinking water disinfection by-products, Ozone Sci. Eng., 14 (1992) 51-69. S.S. Madaeni, Investigation of the mechanism of [51 critical flux in membrane filtration using electron microscopy, J. Porous Materials, 4 (1997) 239-244. [61 B. Schlichter, Entwicklung und Untersuchung eines neuartigen Wasseraufbereitungsverfahrens bestehend aus Wasserozonung und Mikro-/ Ultrafiltration, Mensch & Buch Verlag, Berlin, 2003 [in press]. H. De Balmann and V, Sanchez, Coupling between [71 concentration polarisation and fouling in ultrafiltration, Intnl. J. Chem. Eng., 4(34) (1992) 665673. [II