-
Journal of Membrane Science 289 (2007) 231240
Pre-treatments to reduce foulingme
hririk
ment,, SA 5hemi
SW 2ber 2
r 200
Abstract
Despite the fact that natural organic matter (NOM) is
significantly smaller than the pore size of microfiltration (MF)
membranes, fouling dueto organic compounds has emerged as a
problematic issue for both potable water and wastewater membrane
filtration. Pre-treatment of the feedwater can be a useful strategy
for reduction or mitigation of these fouling effects. The aim of
this investigation was to evaluate various combinedpre-treatment
methods for reducing NOM fouling of laboratory scale
micro-filtration (MF) membranes, including treatment with
adsorbents suchas MIEX (MIEX is a registered trademark of Orica
Australia Pty. Ltd.) and powdered activated carbon (PAC) as well as
coagulation withalum. Highbe a simpleexamined. Rincluding cbut did
not r 2007 Else
Keywords: M
1. Introdu
Foulingnificant isspotable antration hasto foulingin the
fieldthe complelack of undthe need fopossible m
Microfiltypically u
CorresponE-mail ad
0376-7388/$doi:10.1016/jperformance size-exclusion
chromatography (HPSEC) was applied to determine molecular weight
(MW) distribution and proved toanalytical technique, capable of
detecting the onset of fouling by observation of the >50,000 Da
colloidal peak in the water sourcesesults obtained showed that
treatments that reduce the majority of bulk water dissolved organic
carbon (DOC) of all MW ranges,
olloidal (very high MW) material successfully prevented
short-term fouling of MF, where treatments that removed most of the
DOCemove the colloidal components, were unable to prevent
fouling.vier B.V. All rights reserved.
icrofiltration; MIEX; NOM; PAC; Fouling
ction
of membranes by natural organic matter is a sig-ue for the
efficiency of membrane filtration in bothd wastewater treatment
systems. As membrane fil-grown in prevalence, the issue of flux
decline due
has determined the direction of significant research. While
inorganic fouling is as much an issue, it isx and often unknown
composition of NOM and aerstanding of the fouling mechanisms that
has drivenr scientific investigation to address the causes and
eans of mitigation.tration (MF, pore sizes between 0.1 and 10m)
issed as a clarification process for the removal of par-
ding author. Tel.: +61 8 8259 0314; fax: +61 8 8259 0228.dress:
[email protected] (R. Fabris).
ticulate material. This may be as a polishing step following
aconventional treatment or as a pre-treatment before a more
reten-tive membrane process such as nanofiltration (NF) or
reverseosmosis (RO). It is worth noting that dissolved organic
car-bon (DOC) is not typically retained by MF due to the poresizes
involved being much larger than component molecules;however DOC is
nevertheless involved in both short and longterm fouling. Research
has shown that only a small portion ofthe total NOM is responsible
for irreversible fouling includinghigh molecular weight (MW)
polysaccharides, colloidal mate-rial, low MW proteins and amino
sugars [14], however highlyaromatic hydrophobic acids that make up
the majority of typicalnatural water NOM also cause significant
flux decline throughreversible fouling [5]. Recent research has
highlighted the influ-ence of organic colloids in the fouling of MF
membranes [610]and shows that of the fractional components of NOM
in a watersource, the colloids cause the most significant flux
decline [9].Colloidal fouling also appears independent of solution
pH but
see front matter 2007 Elsevier B.V. All rights
reserved..memsci.2006.12.003micro-filtration (MF)Rolando Fabris a,,
Eun Kyung Lee b, C
Vicki Chen b, Mary Da Co-operative Research Centre for Water
Quality and Treat
SA Water Corporation, PMB 3, Salisburyb UNESCO Centre for
Membrane Science and Technology, School of C
University of New South Wales Sydney, NReceived 20 April 2006;
received in revised form 27 Novem
Available online 8 Decembeof low pressurembranes
stopher W.K. Chow a,as aAustralian Water Quality Centre,110,
Australiacal Engineering and Industrial Chemistry,052,
Australia006; accepted 1 December 20066
-
232 R. Fabris et al. / Journal of Membrane Science 289 (2007)
231240
increases in the presence of divalent metal ions such as
calcium[9,10].
Strategiful membraadsorption)onset of fotarget dailyface area
(iplant footpcycles andexpense inenvironmeMembranecapacity foity.
Many da hydrophiinteractionthe membrand also thhaving mucbranes
with
Pre-treamodify potion of
flumembranetreatmentlation/floccfiltration odirectly onon the
surfa[1416]. Itreduces thetrant NOMet al. [4]
reirreversibleadditionalemployed,there is extNOM levemembranetion
[1722resins [23]the membrtime to fultion, the acmay createfouling
[25tion with adadd to an e
The aimtreatment mfor potableusing magnusing alumand variedDOC
overbeen show
pounds [2628]. While most means of evaluating membranefouling
are destructive or require the removal of the mem-
from the system, it was hoped that the contributions ofs organic
carbon species to the fouling could be eval-using high performance
size exclusion chromatographyC) applied to the process stream
waters.
terials and methods
ource waters
rce waters chosen for this investigation were both wellterised
by the authors and also known to cause significant
without treatment. Source waters were also selected toe low and
high range levels of DOC to assess the impact onatment and membrane
fouling. The Myponga Reservoirted about 50 km south of Adelaide,
Australia. The wateryponga Reservoir is sourced via surrounding
catchment
genezen
in co3 HUent
ys s
re-tr
com
tep temiagn
exchM inle egan
R2,eatedmin.pore
fines
ig. 1.es to reduce fouling can include reducing flux, care-ne
material selection and pre-treatment (coagulation,. Reduction of
the flux can in some cases prevent theuling altogether, however in
an application where aflux is required this will necessitate
additional sur-
.e. membrane modules) thereby increasing both therint and
capital cost. The benefits of longer filtrationless chemical
cleaning may not justify the additionala purely economic sense but
may become viable if
ntal sustainability is an important consideration [11].materials
will have a significant effect on the foulingr any particular
compound as well as the reversibil-ifferent polymer membranes are
available with eitherlic or hydrophobic surface, thereby changing
theirwith potential foulants [1,12]. The microstructure ofane will
determine the uniformity of the pore sizee resistance to fouling
with track-etched membranesh greater resistance to fouling than
sponge-like mem-large pore openings and irregular pore size
[13].
tment of the process stream to either remove ortential foulants
is an effective method for reduc-x decline and is usually easy to
implement wherefiltration is retrofitted into an existing
conventionalplant. The most popular pre-treatment is coagu-ulation
followed by either traditional rapid sandr direct filtration, where
flocculated water is flowedto the membrane, forming a porous, low
density cakece that is easily removed by scouring or backwashingis
generally accepted that reduction of overall DOCpotential for
fouling, however remaining recalci-
may still be available to foul the membrane. Kimuramarked that
pre-coagulation alone does not mitigate
fouling, only reversible fouling. To remove theserecalcitrant
components, other technologies must besuch as oxidative or
absorptive processes. Whileensive literature on the use of
adsorbents to reducels, studies of the effects of adsorption
treatments onfouling examine mostly activated carbon applica-] and
little on other adsorbents such as ion-exchange
. In most cases, the carbon is applied directly withinane
reactor and may not provide sufficient contactly exploit the
capacity of the carbon [24]. In addi-cumulation of the carbon on
the membrane surfacedifficulties in partitioning the sources of any
resultant]. Ease of implementation may also be a considera-sorbents
such as activated carbon being the easiest to
xisting coagulation plant with minimal modification.of this
investigation was to evaluate various pre-
ethods for reducing NOM fouling of MF membraneswater treatment,
focussed on combining adsorptionetic ion exchange resin (MIEX) with
coagulationor adsorption with powdered activated carbon
(PAC)combinations of all three. MIEX typically removesa broad range
of molecular weights while alum hasn to be effective for higher
molecular weight com-
branevariouuated(HPSE
2. Ma
2.1. S
Soucharacfoulingprovidpre-treis locafrom Mand is(62 HaDam,water
(catchmSydne
2.2. P
Thethree-smal ch[29]. Manionof NOavailabpaddle(B-KEwas trfor
2011mresin
Frally considered a high colour and high DOC sourceunits (HU)
and 11.7 mg/L, respectively). Woronorantrast, is considered a low
colour, low DOC source
and 2.2 mg/L, respectively) and is sourced from thearea of the
Woronora River, serving the residents ofouthern suburbs, NSW,
Australia.
eatments
bined treatments protocol (Fig. 1) was based on areatment
utilising adsorbent technologies and mini-cal addition that was
developed in a previous studyetic ion exchange resin (MIEX) is a
macroporousange resin specifically developed for the removal
drinking water treatment. Detailed description islsewhere
[3032]. Jar tests were performed on a six-g stirrer (SEM Pty. Ltd.,
Australia) in 2 l gator jarsPhipps & Bird, USA). Myponga
Reservoir water
with 10 mL/L of MIEX by stirring at 100 rpmSettled water was
decanted and filtered through ansize filter (Whatman International,
UK) to remove
. The filtered water was contacted with 40 mg/L
Pre-treatment protocol for membrane fouling experiments.
-
R. Fabris et al. / Journal of Membrane Science 289 (2007) 231240
233
of a coal-based, steam-activated powdered activated carbon(PAC)
(PICA, Australia) by stirring at 100 rpm for 30 min. A20,000 mg/L
Al2(SO4)318H2O solution was used to dose coag-ulant at 40 mg/L to
flocculate the PAC and/or the remainingnatural water turbidity. For
coagulation treatment, samples wereflash mixed at 200 rpm for 1 min
followed by 14 min of slowmixing at 25 rpm and 15 min of settling
before samples weregravity filtered through 11m pore size paper
filters (Grade 1,Whatman International Ltd., UK). MIEX contacted
water wasalso coagulated without prior PAC treatment. For all
combinedtreatments, the treated water was filtered through an 11m
poresize filter to remove remaining MIEX and/or PAC,
therebyminimising any further adsorption before the membrane
experi-ments could commence. Samples were taken at the
intermediatetreatment steps to enable partitioning of the
contribution to thefouling of remaining NOM components.
2.3. Membrane congurations
A flat sheet and a hollow fibre submerged configuration
wereutilized in the experiments. Schematics of the filtration
set-up are
shown in Fig. 2. The flat sheet configuration consisted of a
dead-end cell containing a 0.22m pore size hydrophilic flat
PVDFmembrane (GVWP from Millipore) with 15.2 cm2 area. The cellwas
operated in unstirred mode at 30 kPa. The pure water fluxfor the
flat sheet membrane at 30 kPa was 1974 59.3 L/(m2 h)(65.8 1.98
L/(m2 h kPa)). The submerged hollow fibre (SHF)module consisted of
10 fibres (30 cm length) of 0.2m poresize hydrophilic PVDF (US
Filter), potted in-house. The lowerend of the bundle was fixed and
blocked while suction wasapplied in the lumen of the fibres from
the top of the bun-dle. At 100 L/(m2 h), the measured transmembrane
pressure(TMP) was 0.62 0.02 kPa for pure water (161.3 L/(m2 h
kPa)).In both configurations, the TMP and flux were monitored
usinga pressure transducer and an electronic balance connected to
acomputer.
2.4. Instrumental analyses
Analysed parameters included turbidity, true colour,
ultra-violet absorbance at 254 nm (UV254), DOC, molecular
weightdistribution by high performance size exclusion
chromatography
Fig. d (b) s2. Schematic of (a) dead-end unstirred filtration
set-up in constant pressure an ubmerged hollow fibre filtration
set-up in constant flux.
-
234 R. Fabris et al. / Journal of Membrane Science 289 (2007)
231240
(HPSEC) and scanning electron microscopy (SEM). Turbid-ity was
determined using a Hach 2100AN turbidimeter (Hach,USA) and is
expressed in nephelometric turbidity units (NTU).Samples for true
colour, UV254 and DOC were filtered through0.45m membranes. True
colour was measured using a 5 cmquartz cell at 456 nm and
calibrated against a Platinum/Cobaltstandard [33]. UV254 was
measured through a 1 cm quartz celland DOC was measured using a
Sievers 820 Portable TOCanalyser (Ionics, USA). HPSEC was analysed
using a WatersAlliance 2690 separations module and 996 photodiode
arraydetector (PDA) at 260 nm (Waters Corporation, USA). Phos-phate
buffer (0.1 M) with 1.0 M NaCl was flowed through aShodex KW802.5
packed silica column (Showa Denko, Japan)at 1.0 mL/min. This column
provides an effective separationrange from approximately 100 Da to
an exclusion limit of50,000 Da. Apparent molecular weight was
derived by cali-bration witstandards operformedmicroscope
2.5. Memb
The fodescribed bDarcys lawseries mod
J = (Rm
where J is tpermeate vtotal foulan(Rrev) and iRf = RrevThe
resistaing pure wthen changmonitoredfiltration cein situ
threcentration
only, while avoiding significant disturbance of the NOM
foulinglayer. The pure water flux of the membrane was then
re-evaluatedfor the determination of irreversible foulant
resistance. Rf treat isdefined as the calculated total foulant
resistance of the treatedwater.
3. Results and discussion
3.1. Pre-treated water quality
When comparing water quality data for the various pre-treatments
(Table 1), it is clear that MIEX was very effectiveas a primary
treatment for colour, UV absorbance and DOCremoval. Although
initial water turbidity was not high foreither water source,
treatments that reduced turbidity to around0.1 NTU in both source
waters included alum coagulation, as
sorbe ca
wa
1) indue
sorbe. Cpre-as n
ct orof th
. Fuan 1.(>80diffeetersen s
embrvery
seen
d tomplical
indts si
e notnded
Table 1Treated water
Sample
Myponga RawMyponga MIMyponga MIMyponga MIMyponga MI
Woronora RaWoronora MIWoronora MIWoronora MIWoronora MIh
poly-styrene sulphonate (PSS) molecular weightf 35, 18, 8 and 4.6
kDa. Electron microscopy was
using a Hitachi S900 field emission scanning electron(Hitachi
Science Systems, Japan).
rane fouling
uling behaviour of membrane filtration can bey the
resistance-in-series model, which is based on. For constant
pressure filtration, the resistance-in-
el is expressed as:P
+ Rf) (1)
he filtrate flux, P the transmembrane pressure, theiscosity, Rm
the membrane resistance, and Rf is thet resistance. Rf is the sum
of hydraulically reversiblerreversible (Rirrev) fouling
resistances:+ Rirrev (2)nce of a clean membrane was determined by
filter-ater until steady state was attained. The feed wased to
pre-treated water and the filtrate flow rate wasin order to
determine Rf. At the end of filtration, thell was emptied and the
membrane was gently rinsed
e times with 50 mL of pure water to remove the con-polarisation
layer contributing to reversible fouling
the abin somtreated(Tableinantlythe adfeasiblto thetion win
direeffectsbranesless thwaterlargeparamhad beMF mties
ofpeak,believeand cobiologusuallyprevenand aror exte
quality parameters for pre-treated Myponga Reservoir and
Woronora Dam
Code pH Turbidity (NTU)MR 7.8 1.95
EX MM 6.9 0.66EX/Alum MMA 6.5 0.12EX/PAC MMP 7.3 0.24EX/PAC/Alum
MMPA 6.6 0.14
w WR 6.7 0.69EX WM 6.5 0.28EX/Alum WMA 4.8 0.07EX/PAC WMP 6.6
0.25EX/PAC/Alum WMPA 4.7 0.07ents are largely incapable of
turbidity removal andses (especially PAC) increase the visually
apparentter turbidity prior to filtration. Reported reductions
turbidity for the adsorbent treatments were predom-to the
subsequent filtration step, as partitioning of
ents and natural water turbidity was not practicallyoarse (11m)
filtration was also deemed necessarytreatment procedures as the
focus of the investiga-ot to observe the effects of particles and
DOC, ascontact microfiltration, but rather the more isolatede
dissolved organic species on fouling of MF mem-
ll combined treatment produced treated water with0 mg/L DOC for
both Myponga and Woronora source% DOC removed). It is worth noting
that despiterences in both DOC and traditional water quality, both
Myponga Reservoir and Woronora Dam waterhown to produce significant
short term fouling ofanes. Both waters also contained detectable
quanti-high MW colloidal NOM. This multi-component
at the exclusion limit of the column (50,000 Da), isbe composed
of some NOMmetal complexes [28]
ex amino sugars from bacterial cell walls and othersources
[34,35]. The organo-metallic complexes areicative of low residence
time in the catchment, whichgnificant natural photo-oxidation or
biodegradation,generally detected following conventional
treatmentstorage. The amino sugar component however, has
Colour (HU) UV254 (cm1) DOC (mg/L)65 0.432 11.76 0.054 3.41
0.037 2.82 0.013 1.00 0.007 0.9
3 0.035 2.21 0.013 1.22 0.021 1.11 0.005 0.40 0.005 0.4
-
R. Fabris et al. / Journal of Membrane Science 289 (2007) 231240
235
Fig. 3. Pre-treatment DOC molecular weight distributions by
HPSEC for Myponga Reservoir. Code: R, Raw; M, MIEX; A, Alum; P,
PAC.
generally proved to be more difficult to remove. For the
purposesof this papthe sourcegreater than
MIEXcomponentDOC (1000colloidaling high Mremoved loafter
MIEXically it isremoval oftheir inabilstructure ddrance on t
3.2. Flat sheet fouling experiments
fouin F
, Mymbrent o
ancmayeatmNOMater.d infou
avetio (Rer, colloidal material is defined as the component
ofwater organic matter of apparent molecular weight
50,000 Da that is detectable by UV absorbance.treatment reduced
a broad range of UV-absorbing
s (Figs. 3 and 4), especially high and medium MW10,000 Da) but
was not effective for very high MWNOM. Coagulation with alum
removed the remain-W NOM and all detectable colloidal material.
PACw to medium MW DOC (3001000 Da) when used contact but little
additional colloidal material. Typ-observed that most absorbents
are ineffective forcolloidal components. This could be explained
by
ity to take dissolved colloidal material into the poreue to
physical size, as well as possible stearic hin-he adsorbent
surface.
Theshownwatersthe metreatmperformrangepost-trdenseraw w
resultegreatermay hThe raFig. 4. Pre-treatment DOC molecular
weight distributions by HPSEC for Woronoralant resistance as a
function of permeate volume isig. 5 for raw and treated waters. For
the untreatedponga water being higher in colour and DOC foulsane
more readily than Woronora water. The MIEXf Woronora water resulted
in detrimental filtratione. The broken MIEX resin fines in the
sub-micronnot have been completely removed by the 11ment filtration
and interacted with NOM to form a-MIEX cake layer, resulting in the
higher Rf thanThe MIEX treated Myponga water, in contrast,reduced
fouling compared to the raw water. The
lant load with regards to the untreated Myponga watermasked any
additional fouling from MIEX fines.
f treat/Rf) is plotted against (Rf) and summarizes theDam. Code:
R, Raw; M, MIEX; A, Alum; P, PAC.
-
236 R. Fabris et al. / Journal of Membrane Science 289 (2007)
231240
Fig. 5. Foulinand treated MDam water. C
benefits ofresistancethe total foof pre-treatthe relative
Compardistributionon molecuthat encomshown). Thsmall to
beexaminatiodecrease wthe designative of spec
Fig. 6. Foulaand WoronoraA, Alum; P, P
and retentate samples were collected at the termination of
theexperiment and therefore represent averaged water quality.
Thismeans that they contain material from both before, and after
theonset of fouling. In interpretation of the fouling behaviour of
thevarious pre-treated waters, observation of the molecular
weightdistribution would tend to propose two fouling theories.
Eitherthe colloidal material is directly involved in the membrane
foul-ing and therefore becomes less abundant in the permeate
withincreasing volume filtered, or the membrane fouling by
othercomponents causes increased retention of the colloidal
mate-rial. Both theories are valid if based only on the HPSEC
data,however TMP increases were only apparent in treated
waterswhere the colloidal material was still present indicating
that thecolloidal DOC was directly involved in the short-term
foulingobserved. This further confirms the findings of Lee et al.
andChen et al. [6,36]. The data presented in Figs. 7 and 8 also
showthat in most cases, the amount of colloidal DOC detected inthe
permea
on c
the csolual Dmbr
lk aqtersd mal Dter fts ared aIEXed thlesseg profile (foulant
resistance) vs. permeate volume of (a) untreatedyponga Reservoir
water and (b) untreated and treated Woronoraode: R, Raw; M, MIEX;
A, Alum; P, PAC.
multiple pre-treatment in terms of reduced foulantin Fig. 6. For
this investigation, Rf treat is defined asulant resistance at a
permeated volume of 1000 mLed water. A ratio (Rf treat/Rf) of less
than 1.0 indicatesbenefit of treatment in terms of reduced
fouling.
ison with the observations of the molecular weightby HPSEC
showed that MF alone had no effect
lar weight components in the 3008000 Da range
nomen
Whileout thecolloidthe methe bu
Wareducecolloidface affoulanprovidand Mobservdue topassed the
majority of the UV absorbance (data notis was entirely expected as
these components are tooretained by a 0.2m pore size membrane.
However,n of the >50,000 Da response consistently showed aith
increasing permeate volume (Figs. 7 and 8). Whileted 500 and 1000
mL samples were both representa-ific time periods during the
filtration, the permeate
nt resistance benefit analysis of pre-treatments for Myponga
(M)(W) water (ratio < 1.0 is a benefit). Code: R, Raw; M,
MIEX;
AC.
ments thatcan also bedata and thon microfiremove retaof
foulinghowever thnal resistandecrease in
Table 2Percentages o
Sample
Myponga RawMyponga MIMyponga MIMyponga MIMyponga MI
Woronora RaWoronora MIWoronora MIWoronora MIWoronora MI
Rf, foulant reste exceeded the amount in the retentate. This
phe-an also be explained by the proposed fouling theory.olloidal
DOC in the permeate is distributed through-tion, following the
onset of fouling, the majority ofOC in the retentate is immobilised
on the surface ofane (fouling layer) and is therefore less abundant
inueous solution.pre-treated with MIEX + alum showed greatlyembrane
fouling in parallel with high removal ofOC. This is supported by
SEM of the membrane sur-ouling and washing (Fig. 9) which shows the
surfacee clearly reduced with pre-treatment and that
alumsignificant improvement. In comparing coagulation
pre-treatment for ultrafiltration, Son et al. [37] alsoat MIEX
was less effective for reduction of foulingr removal of high MW
organic matter. For the treat-included PAC, some residual
particulates from PACseen which is consistent with the obtained
resistancee work of Matsui et al. [24] with submicrometre
PACltration. After washing of the membrane surface toined material
and the filter cake (Table 2), the amountmaterial in absolute terms
was clearly reduced,
e reversible component as a percentage of the origi-ce decreased
with pre-treatments despite an overalltotal resistance. This
suggests that although several
f reversible resistance after flat sheet filtration
Code Rf treat (1/m) %Rf reversibleMR 1.15 1012 60.9
EX MM 5.16 1011 46.4EX/Alum MMA 4.20 109 44.6EX/PAC MMP 1.05
1011 53.9EX/PAC/Alum MMPA 7.22 109 21.5w WR 4.68 1011 44.8EX WM
6.73 1011 42.9EX/Alum WMA 7.71 109 72.6EX/PAC WMP 1.55 1011
31.6EX/PAC/Alum WMPA 8.74 109 33.9istance; Rf treat, foulant
resistance at 1000 mL permeated volume.
-
R. Fabris et al. / Journal of Membrane Science 289 (2007) 231240
237
Fig. 7. Impactreated water
pre-treatmesurface foublocking co
Treatmeshowed hi(Table 2). Tefit of remmay be ovin short
tersource likefines on theof DOC dupre-treatme
3.3. Subme
For the sthe waters tthe MIEXwaters. It wthe successt of
flat-sheet micro-filtration fouling on retention of high molecular
weight colloidaland (c) Myponga MIEX and PAC treated water.
nts were very effective for reduction of short term,ling (cake
formation), there still remains some porenstituents that can
contribute to longer term fouling.nt of both water sources with
MIEX + PAC + alumgher Rf treat than MIEX + alum treatments alonehis
indicates that at high levels of treatment, the ben-
oving additional low MW organic material by PACerweighed by the
detrimental effects of PAC finesm fouling of MF. Similarly, for a
low DOC waterWoronora, the detrimental effect of MIEX resinfouling
may be more significant than the reduction
e to the treatment. This highlights the need to tailornt
strategies for an individual water source.
rged hollow bre experiments
ubmerged hollow fibre experiments at 100 L/(m2 h),hat were
analysed were both untreated source waters,
+ alum treated and MIEX + PAC + alum treatedas intended that the
SHF experiments would test if
ful treatments from the flat-sheet experiments would
also prevenfluxes. Pre-the flat-shewater condfor the durtreated
watlated filtratshown in Ffiltration wat constantthe foulinghigher
fluximately 30more foulin
As a resthe appliedthan those asource watthe plannedincreases
into the onsetNOM using (a) Myponga Reservoir water, (b) Myponga
MIEX
t fouling in a more practical application and at
highertreatments that failed to prevent short term fouling inet
experiments were not applied. For the two treateditions, the set
flux was maintained at the same TMPation of the filtration (2.5 L,
4 h), indicating that theers failed to foul the hollow fibre
bundle. The accumu-ion resistances at a permeated volume of 1000 mL
areig. 10. The observed fouling behaviour during SHFas similar to
that of dead-end unstirred cell filtrationpressure shown in the
previous section. However,resistances were less. One reason may be
that the
es obtained with constant pressure filtration (approx-0 L/(m2
h)) in the flat sheet configuration producedg.
ult, it can be assumed that the low fouling potential oftreated
waters extended to volumes and fluxes greaterpplied to the
flat-sheet fouling experiments. The raw
er experiments were both terminated after 1.5 L of2.5 L total
volume of water was filtered as appliedTMP were unable to maintain
the desired flux due
of fouling. HPSEC scans of the colloidal peak reveal
-
238 R. Fabris et al. / Journal of Membrane Science 289 (2007)
231240
Fig. 8. Impact of flat-sheet micro-filtration fouling on
retention of high molecular weight colloidal NOM using (a) Woronora
Dam water, (b) Woronora MIEXtreated water and (c) Woronora MIEX and
PAC treated water.
Fig. 9. Electron microscopy of Woronora (W) and Myponga (M)
flat-sheet membrane after fouling and rinsing. Code: R, Raw; M,
MIEX; A, Alum; P, PAC.
-
R. Fabris et al. / Journal of Membrane Science 289 (2007) 231240
239
Fig. 10. Tran (a) MM, MIEX; A
Fig. 11. ImpaWoronora Da
that by therial was verpermeate (MIEX + Ping of micrincreased
tsheet experindicate thadditional r
4. Conclu
In the apexpected thorganic maeffects ofbe more ereduced
thsmembrane pressure (TMP) during submerged hollow fibre (SHF)
filtration of, Alum; P, PAC.ct of submerged hollow fibre
microfiltration fouling on retention of high molecular wm
water.
1000 mL sampling point, retention of colloidal mate-y high in
both raw water sources with low levels in theFig. 11). The
pre-treatment with MIEX + alum andAC + alum, successfully prevented
short-term foul-oporous SHF. The addition of PAC treatment
slightlyhe fouling resistance as was also observed in the
flatiments; however, the SEMs of flat sheet membranesat the
presence of PAC fines may contribute to thisesistance.
sion
plication of the combined treatment protocol, it wasat the
removal of selective components of the sourcetter by the various
treatment steps would allow thethe residual organic material on
fouling of MF toasily observed. It was shown that treatments thate
majority of bulk water DOC of all MW ranges,
including cMIEX + Ping of MFDOC (MIEponents, wa
non-destrsignificantthe >50,000examined.be used excability to
aout removaa monitorin
Reference
[1] L. Fan, Jistics of nWater Reyponga Reservoir and (b)
Woronora Dam water. Code: R, Raw;eight colloidal NOM using (a)
Myponga Reservoir water and (b)
olloidal (very high MW) material (MIEX + Alum,AC + Alum)
successfully prevented short-term foul-, where as treatments that
removed most of theX + PAC) but did not remove the colloidal
com-
ere unable to prevent fouling. HPSEC proved to beuctive
analytical technique, capable of detecting acontributor to membrane
fouling by observation ofDa colloidal organic material in the water
sources
Although it cannot provide sufficient information tolusively in
the evaluation of membrane fouling, the
pply HPSEC analysis to an operating system with-l of the
membrane makes it potentially attractive asg method.
s
.L. Harris, F.A. Roddick, N.A. Booker, Influence of the
character-atural organic matter on the fouling of microfiltration
membranes,s. 35 (18) (2001) 44554463.
-
240 R. Fabris et al. / Journal of Membrane Science 289 (2007)
231240
[2] B. Kwon, S. Lee, J. Cho, H. Ahn, D. Lee, H.S. Shin,
BiodegradabilityDBP formation and membrane fouling potential of
natural organic matter:characterisation and controllability,
Environ. Sci. Technol. 39 (3) (2005)732739.
[3] N. Her, G. Amy, H.R. Park, M. Song, Characterizing algogenic
organicmatter (AOM) and evaluating associated NF membrane fouling,
Water Res.38 (6) (2004) 14271438.
[4] K. Kimura, Y. Hane, Y. Watanabe, Effect of pre-coagulation
on mitigat-ing irreversible fouling during ultrafiltration of a
surface water, Water Sci.Technol. 51 (67) (2005) 93100.
[5] L. Fan, J. Harris, F. Roddick, N. Booker, Fouling of
microfiltration mem-branes by the fractional components of natural
organic matter in surfacewater, Water Sci. Technol.: Water Supply 2
(56) (2002) 313320.
[6] V. Chen,membration and
[7] A.I. Schmembraon Memb2000, De2, pp. 41
[8] B. Kwonfouling apolysaccTechnol.
[9] N. Lee,associateWater R
[10] A.R. Cosorganic716725
[11] K. ParamfiltrationWater R
[12] T. Thorsin NF an
[13] H.H.P. Fsludge, J
[14] T. Leikn(NOM)metal me
[15] M.-H. Cpermeab(2005) 1
[16] A.W. Zuin memb194 (13
[17] M. Pirbabrane fil269127
[18] K. Konietreatmen
[19] C.-W. Limoleculation 170
[20] S. Moziaacids andRes. 39 (
[21] P. Zhao, S. Takizawa, H. Katayama, S. Ohgaki, Factors
causing PACcake fouling in PAC-MF (powdered activated
carbon-microfiltration) watertreatment systems, Water Sci. Technol.
51 (67) (2005) 231240.
[22] R. Thiruvenkatachari, W.-G. Shim, J.-W. Lee, R.B. Aim, H.
Moon, A novelmethod of powdered activated carbon (PAC) pre-coated
microfiltration(MF) hollow fiber hybrid membrane for domestic
wastewater treatment,Colloids Surf. A: Physicochem. Eng. Aspects
274 (13) (2006) 2433.
[23] S. Khirani, P.J. Smith, M.-H. Manero, R.B. Aim, S.
Vigneswaran,Effect of periodic backwash in the submerged membrane
adsorptionhybrid (SMAHS) for wastewater treatment, Desalination 191
(13) (2006)2734.
[24] Y. Matsui, R. Murase, T. Sanogawa, N. Aoki, S. Mima, T.
Inoue, T.Matsushita, Rapid adsorption pretreatment with
submicrometre powdered
ivated7) (20N. Jaim, E
a micter Su.K. C
e, ThatmenDrikracterowthgulat. Allp
Brinkize do(2005Fabriural onferenM anChowIEX
bact274. Singin, W. Moratmenpply 4. Ben
s. 27 (. Leectivityter ScMakdfonehnol.Lee, GfoulinOM),.
SonplicatNOM5) (20A.G. Fane, S. Madaeni, I.G. Wenten, Particle
deposition duringne filtration of colloids: transition between
concentration polariza-cake formation, J. Membr. Sci. 125 (1)
(1997) 109122.afer, A.G. Fane, T.D. Waite, Fouling effects on
rejection in thene filtration of natural waters, in: Proceedings of
the Conferenceranes in Drinking and Industrial Water Production,
vol. 1, Octobersalination Publications, LAquila, Italy, 2000, ISBN
0-86689-060-1420., S. Lee, M.B. Gu, J. Cho, Minimisation of
membrane organicnd haloacetic acids formation by controlling amino
sugars and/orharide-like substances included in colloidal NOM,
Water Sci.: Water Supply 3 (5) (2003) 223228.G. Amy, J.-P. Croue,
Low-pressure membrane (MF/UF) foulingd with allochthonous versus
autochthonous natural organic matter,
es. 40 (12) (2006) 23572368.ta, M.N. de Pinho, M. Elimelech,
Mechanisms of colloidal natural
matter fouling in ultrafiltration, J. Membr. Sci. 281 (12)
(2006).eshwaran, A.G. Fane, B.D. Cho, K.J. Kim, Analysis of
micro-performance with constant flux processing of secondary
effluent,
es. 35 (18) (2001) 43494358.en, Concentration polarisation by
natural organic matter (NOM)d UF, J. Membr. Sci. 233 (12) (2004)
7991.ang, X. Shi, Pore fouling of microfiltration membranes by
activated. Membr. Sci. 264 (12) (2005) 161166.es, H. Odegaard, H.
Myklebust, Removal of natural organic matterin drinking water
treatment by coagulation-microfiltration usingmbranes, J. Membr.
Sci. 242 (12) (2004) 4755.ho, C.-H. Lee, S. Lee, Influence of floc
structure on membraneility in the coagulation-MF process, Water
Sci. Technol. 51 (67)43150.larisam, A.F. Ismail, R. Salim,
Behaviours of natural organic matterrane filtration for surface
water treatmenta review, Desalination) (2006) 211231.zari, V.
Ravindran, B.N. Badriyha, S.-H. Kim, Hybrid mem-
tration process for leachate treatment, Water Res. 30 (11)
(1996)06.czny, G. Klomfas, Using activated carbon to improve
natural watert by porous membranes, Desalination 147 (1) (2002)
109116., Y.-S. Chen, Fouling of membrane by humic substance: effect
ofr weight and powder-activated carbon (PAC) treatment,
Desalina-(1) (2004) 5967., M. Tomaszewska, A.W. Morawski, Studies
on the effect of humicphenol on adsorption-ultrafiltration process
performance, Water
23) (2005) 501509.
act(6
[25] H.-S. KofWa
[26] C.WPagtre
[27] M.charegcoa
[28] B.PT.ter(7)
[29] R.natCoRO
[30] C.(Mand267
[31] P.Cres
[32] J.YtreSu
[33] L.ERe
[34] J.Area
Wa[35] G.
sulTec
[36] N.of(N
[37] H.JApfor5 (carbon particles before microfiltration, Water
Sci. Technol. 5105) 249256.
ng, D.-S. Lee, M.-K. Park, S.-Y. Moon, S.-Y. Cho, C.-H. Kim,
H.-ffects of the filtration flux and pre-treatments on the
performance
rofiltration drinking water treatment system, Water Sci.
Technol.:pply 6 (4) (2006) 8187.how, J.A. van Leeuwen, M. Drikas,
R. Fabris, K.M. Spark, D.W.
e impact of the character of natural organic matter in
conventionalt with alum, Water Sci. Technol. 40 (9) (1999)
97104.as, C.W.K. Chow, D. Cook, The impact of recalcitrant organicr
on disinfection stability, trihalomethane formation and bacterial:
an evaluation of magnetic ion exchange resin (MIEX) and alumion, J.
Water Supply Res. Technol. Aqua 52 (7) (2003) 475487.ike, A. Heitz,
C.A. Joll, R.I. Kagi, G. Abbt-Braun, F.H. Frimmel,
mann, N. Her, G. Amy, Size exclusion chromatography to charac-c
removal in drinking water treatment, Environ. Sci. Technol. 39)
23342342.
s, C.W.K. Chow, M. Drikas, Combined treatments for
enhancedrganic matter (NOM) removal, in: Proceedings of the Enviro
06ce, Melbourne, Australia, 911th May, 2006, paper e6174, CD-
d www.enviroaust.net/e6/papers/e6174.pdf., D. Cook, M. Drikas,
Evaluation of magnetic ion exchange resin) and alum treatment for
formation of disinfection by-productserial regrowth, Water Sci.
Technol.: Water Supply 2 (3) (2002).er, K. Bilyk, Enhanced
coagulation using a magnetic ion exchange
ater Res. 36 (16) (2002) 40094022.ran, M. Drikas, D. Cook, D.B.
Bursill, Comparison of MIEXt and coagulation on NOM character,
Water Sci. Technol.: Water(4) (2004) 129137.
nett, M. Drikas, The evaluation of colour in natural waters,
Water7) (1993) 12091218.nheer, Comprehensive assessment of
precursors, diagenesis, and
to water treatment of dissolved and colloidal organic matter,i.
Technol.: Water Supply 4 (4) (2004) 19.issy, J.-P. Croue, G. Amy,
H. Buisson, Fouling of a polyether-
ultrafiltration membrane by natural organic matter, Water Sci.:
Water Supply 4 (4) (2004) 205212.. Amy, J.-P. Croue, H. Buisson,
Identification and understanding
g in low-pressure membrane (MF/UF) by natural organic
matterWater Res. 38 (20) (2004) 45114523., Y.D. Hwang, J.S. Roh,
K.W. Ji, P.S. Sin, C.W. Jung, L.S. Kang,ion of MIEX pre-treatment
for ultrafiltration membrane processremoval and fouling reduction,
Water Sci. Technol.: Water Supply
05) 1524.
Pre-treatments to reduce fouling of low pressure
micro-filtration (MF) membranesIntroductionMaterials and
methodsSource watersPre-treatmentsMembrane
configurationsInstrumental analysesMembrane fouling
Results and discussionPre-treated water qualityFlat sheet
fouling experimentsSubmerged hollow fibre experiments
ConclusionReferences