Desalination 236 (2009) 208–215

A comparative study of microfiltration and ultrafiltration

of activated sludge-lagoon effluent

T. Nguyen, L. Fan, F.A. Roddick*, J.L. Harris

School of Civil, Environmental and Chemical Engineering, RMIT University,
GPO Box 2476V, Melbourne, Australia 3001
email: felicity.roddick@rmit.edu.au
Received 30 June 2007; revised accepted 7 October 2007

Abstract
Melbourne’s Western Treatment Plant is unusual in that it employs a sequential activated sludge-lagoon
(AS-lagoon) system to treat municipal wastewater. Reuse of the treated water is limited for some applications
due to salt content, and membrane pre-treatment prior to reverse osmosis is under consideration. The use of micro-
filtration (MF) and ultrafiltration (UF) for improving the quality of water prior to reverse osmosis was investi-
gated. The organic components of the feed water (AS-lagoon effluent), permeates and foulant layers were
characterised using three-dimensional excitation-emission-matrix (EEM) spectroscopy, attenuated total reflec-
tion-Fourier transform infrared spectroscopy (ATR-FTIR), dissolved organic carbon (DOC) determination, size
exclusion chromatography (SEC) and ultraviolet (UV) absorbance. DOC removal was 28% for UF and 2% for
MF. MF removed mostly non-UV absorbing molecules of AMW 40–70 kDa, whereas UF removed molecules
in this size range, a high proportion of which were UV-absorbing, as well as some organic compounds in the
3–8 kDa range, some of which were UV-absorbing. The organic compounds removed by UF had hydrophobic,
hydrophilic and transphilic character, and were shown to comprise humic-like matter, soluble microbial products
and protein-like extracellular matter. Fulvic-like matter largely passed through the UF membrane. ATR-FTIR
analysis of fouled MF and UF membranes showed that polysaccharides, polysaccharide-like compounds and pro-
teins were the prominent components in the fouling layer.
Keywords: Activated sludge; Lagoon; Microfiltration; Ultrafiltration; Organic fouling; Organic matter

*Corresponding author.

Presented at the International Membrane Science and Technology Conference, IMSTEC 07, 5–9 November 2007,
Sydney, Australia

0011-9164/09/$– See front matter # 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.desal.2007.10.069
 T. Nguyen et al. / Desalination 236 (2009) 208–215
1. Introduction
Western Treatment Plant (WTP) treats
approximately 52% of Melbourne’s sewage, a
total of approximately 485 million L/day,
using a sequential activated sludge-lagoon
(AS-lagoon) process. WTP employs two
AS-lagoon systems in which sewage is treated
by passing through ponds and an activated sludge
plant with anoxic and aeration zones where bio-
degradable matter is consumed by naturally-
occurring bacteria. The biologically treated
effluent then passes through a clarifier and a
chain of lagoons before it goes to the reuse chan-
nel and is transferred to the head of the road sto-
rage pond (HORS). Melbourne Water currently
supplies recycled water from HORS for various
on-site and off-site uses. Due to a number of
catchment issues, including industrial waste
input and saline aquifer infiltration, the effluent
from HORS contains salt which limits the long
term sustainable use of recycled water for some
applications, such as agriculture, in the region
without additional management practices.
Pilot-scale membrane-based processes have
been trialled on-site with the aim of removing
salt from the lagoon effluent. These trials, which
utilised microfiltration (MF) or ultrafiltration
(UF) as a pre-treatment prior to reverse osmosis
(RO), demonstrated that the product water was
suitable for various applications including agri-
culture and domestic use. Since the effluent
from WTP contains algae and algal products
from the lagoon process, as well as some resi-
dual products from the AS process, the resultant
membrane fouling and permeate properties may
be more problematic and differ from that arising
from separate AS and lagoon processes.
Lagoon effluent contains particles, colloids,
salts, natural organic matter (NOM) and soluble
microbial products (SMP) derived from bio-
logical wastewater treatment processes, these
constituents can adsorb and deposit on the mem-
brane surface. Consequently, fouling of the
209
membrane can be due to organic, inorganic,
colloidal and biological matter. Fouling due to
these components can occur simultaneously, and
the components may interact in terms of
mechanism [1], thus the overall process can be
very complex and its mitigation requires careful
evaluation and process development. NOM,
which is ubiquitous in natural waters, is one of
the most recalcitrant among potential foulants
[2,3]. This is due to the heterogeneous nature
of NOM, which unlike other organic macromo-
lecules such as proteins [4], leads to different
fouling mechanisms for different membrane
types [5]. Multivalent cations such as calcium
and magnesium can react with NOM to form
metal–NOM complexes [6–8], changing the
electrokinetic properties of NOM [4] and result-
ing in flux decline due to the formation of a
highly compacted fouling layer [6,9,10].
The extracellular polymeric substances
(EPS) produced by bacteria are considered to aid
bioflocculation and enhance microbial attach-
ment to the membrane surface by cross-linking
and stabilizing the biofilm [11]. Speth et al.
[12] reported that fouling of membrane systems
by EPS was mostly attributed to the deposition
of extracellular material in the cake layer.
Kimura et al. [13] demonstrated that polysac-
charide and polysaccharide-like organic matter
caused irreversible fouling. Although the EPS
have been considered the major contributors to
the gel matrix responsible for the biofouling of
membranes, there is recent evidence that the gel
could also be produced from SMP in the influent
water [14]. The SMP from the biological treat-
ment of wastewater have been shown to be
among the most recalcitrant naturally occurring
organic foulants of UF, NF and RO membranes
[15]. They form a sticky layer on the microbes
which results in the cells adhering to the mem-
brane surface, and each other, causing signifi-
cant irreversible fouling.
A wide variety of colloids is encountered in
wastewaters; they include corrosion products,
210 T. Nguyen et al. / Desalination 236 (2009) 208–215
silt and clay, biological debris, colloidal silica
and sulfur, organic phosphates, and precipitated
iron and aluminium compounds from incom-
plete pre-treatment. Yiantsios and Karabelas
[16] showed that colloid stability played a major
role in UF/RO membrane fouling and that stable
colloidal suspensions caused less membrane
fouling.
Effective pre-treatment is the first line of
defence in controlling membrane fouling and
assuring successful RO operation [17]. The use
of pre-treatment processes such as coagulation–
flocculation, sand filtration, disinfection, ozona-
tion, flotation, activated carbon adsorption and
other membrane processes such as MF and UF
is important to guarantee good and consistent
performance of NF and RO systems [18]. Pre-
treatment by MF or UF is more economically
competitive than conventional methods [19]. The
objective of this study was to characterise the
properties of the organic content of the product
water from MF and UF of the AS-lagoon effluent
from WTP to determine a suitable membrane pre-
treatment prior to RO treatment.
2. Materials and methods
2.1. Feed water
As part of an extended sampling and moni-
toring program, treated water from the final
lagoons and storage pond (HORS) was collected
fortnightly for testing of MF and UF filterability.
The sample collected at HORS on 30 March
2007 was chosen for this study as its character-
istics and filterability were representative of the
samples collected during the monitoring period.
The sample was pre-screened (500 and 130 mm
mesh for MF and UF, respectively). After pre-
screening, the samples were stored at 4 C and
warmed to room temperature immediately
before MF and UF experiments. Testing of sam-
ples was conducted as soon as possible to mini-
mise the effect of storage time. Parameters
measured included turbidity, pH, conductivity,
UV absorbance, total dissolved solids (TDS),
dissolved organic carbon (DOC) and filterability
through both MF and UF membranes. The phy-
sical and chemical characteristics of the feed
water are shown in Table 1.
2.2. MF and UF filtration experiments
An Amicon 8050 stirred cell (effective mem-
brane area of 13.4 cm2) was connected to a feed
reservoir which was operated at a constant
pressure regulated using compressed nitrogen
gas and a magnetic stirrer speed of 430 rpm.
The permeate flux was determined using a
top-loading electronic balance with data logging
function connected to a computer. Millipore
membranes were used: PVDF discs with a nom-
inal pore size of 0.22 mm were used for MF and
PES discs with an average molecular weight
cut-off (NMWCO) of 100 kDa were used for
UF. Pore sizes and materials of these mem-
branes were similar to those employed in the
demonstration trials. The operating pressures for
MF and UF were 70 and 110 kPa, respectively.
2.3. Characterisation of organic matter
Dissolved organic matter was separated into
hydrophobic, hydrophilic, and transphilic frac-
tions using DAX-8 and XAD-4 polymeric
Table 1
Characteristics of AS-lagoon effluent
Parameters Values
pH 7.1
Turbidity (NTU) 2.3
Suspended solids (mg L1) 11
Total dissolved solids (mg L1) 1200
Dissolved organic carbon (mg L1) 12.6
Total algal counts (cells mL1) 11,080
Electrical conductivity (ms cm1) 2100
Hardness, as CaCO3 (mg L1) 200
COD (mg L1) 40
 T. Nguyen et al. / Desalination 236 (2009) 208–215 211

resins. Samples were filtered (0.45 mm cellulose 3000

MF
acetate), acidified to pH 2 with 2 M HCl and 2500

Flux rate (L m–2 h–1)

UF
passed through DAX-8 and XAD-4 resins in 2000
sequence. Organic matter concentrations were
1500
determined using a Sievers 820 TOC analyser
and fraction contents were calculated from the 1000

difference in DOC concentration of column 500

effluents and original sample water.
0
Fluorescence measurements were conducted 0 50 100 150 200 250 300 350 400

Permeate volume (L m–2)

using a Perkin Elmer Luminescence Spectrometer
LS 50B and reported in arbitrary units (AU).
Excitation and emission wavelength ranges
were 200–600 nm. Raw water samples were
membrane filtered (0.45 mm cellulose acetate)
to remove any suspended particulates. ATR-
FTIR analyses of the foulant layer on membrane
surfaces were conducted using a Perkin-Elmer
spectrometer over the wave number range of
650–4000 cm1. Absorbance of the samples at
254 nm (UVA) was measured using a Unicam
UV/Vis spectrophotometer.
The molecular weight distribution of dis-
solved organics in the feed water and permeates
was determined using size exclusion chromato-
graphy with UV detection (SEC-UV) and DOC
detection (SEC-DOC). The SEC analyses were
carried out at Curtin Water Quality Research
Centre, Curtin University, Western Australia.
3. Results and discussion
3.1. MF and UF filterability of the AS-lagoon
Fig. 1. Flux rate for MF and UF of AS-lagoon effluent.
was obtained. Thus in terms of permeate volume
produced, MF would be more attractive than UF
as a pre-treatment for RO.
3.2. DOC removal by MF and UF
DOC removal by MF and UF was 2% and
28%, respectively. The organic fractions of the
feed water, MF and UF permeates are shown
in Fig. 2. The hydrophobic fraction (HPO) was
the most abundant fraction in all three samples
(more than 50%), the transphilic (TPI) and
hydrophilic (HPI) fractions being present in
approximately similar proportions. As little DOC
was removed by MF, although there appeared to
be greatest removal of the HPI fraction, no firm
conclusions can drawn regarding the relative
removals of the fractions when experimental
14

effluent 12 AS-lagoon effluent

MF permeate
Typical curves of flux rate versus permeate 10 UF permeate
DOC (mg L–1)

volume for MF and UF of AS-lagoon effluent are 8

shown in Fig. 1. The MF flux rapidly declined 6
and reached the final flux of 40 L m2 h1 (the
4
desired operating flux rate of the pilot-scale
demonstration plant at WTP) at permeate 2

volume of 466 mL, equivalent to 348 L m2. 0

Total HPO TPI HPI
Rapid initial flux decline also occurred for UF,
a permeate volume of 50 mL (equivalent to Fig. 2. Organic fractions of AS-lagoon effluent and MF
182 L m2) on reaching the desired flux rate and UF permeates.
212 T. Nguyen et al. / Desalination 236 (2009) 208–215
Table 2
Removal of organic fractions by MF and UF
Removal (%)
MF UF
HPO 2 37
TPI <1 11
HPI 5 25
error is taken into account (Table 2). For UF, the
bulk of the DOC retained on the filter belonged
to the HPO (humic substances) fraction, fol-
lowed by the HPI (polysaccharides) fraction and
then the TPI fraction.
3.3. Characterisation of the feed and the
permeate by fluorescence
Excitation-emission matrix (EEM) spectra of
the feed water, MF and UF permeates are shown
in Fig. 3. Four regions of high fluorescence
intensity were observed: peak I at 225/340–
350 nm; peak II at 280–285/340–350 nm, peak
III at 330–350/420–430 nm, and peak IV
(240–250/420–460 nm). Peak I is associated
with protein-like extracellular organic matter
which contains aromatic amino acids, peak II
is attributed to soluble microbial products
Fig. 3. EEM spectra of the feed, MF and UF permeates.
(SMP), peak III contains humic-like fluoro-
phores [20] and peak IV is associated with more
fulvic-like material [21]. The intensity of all
peaks in the MF permeate was not significantly
different from that of the feed, consistent with
the removal of 2% of organic matter (as DOC)
by MF. However, there was a marked reduction
in intensity of peaks I, II, III in the UF permeate
demonstrating that significant amounts of pro-
tein-like extracellular organic matter, SMP and
humic-like fluorophores were removed by the
UF membrane.
3.4. Characterisation of the feed and the
permeate by SEC-UV and SEC-DOC
SEC-UV and SEC-DOC chromatograms of
the feed water, MF and UF permeates are shown
in Fig. 4a and b. The molecular weight distribu-
tions of the UV-absorbing components of the
feed and MF permeate were very similar, with
only a slight reduction for MF in the apparent
molecular weight (AMW) range greater than
60 kDa (Fig. 4a). For the UF permeate there was
some reduction in the 3–8 kDa range and almost
complete removal of UV-absorbing molecules
in the 40–70 kDa range. This is consistent with
the UVA and SUVA results (Table 3), the UVA
values demonstrating that molecules with
 T. Nguyen et al. / Desalination 236 (2009) 208–215 213

20

18 (a) Feed

16 MF permeate
UV absorbance (arbitrary units) 14 UF permeate

12

10

0
100,000 10,000 1000 100
–2

Apparent molecular weight (Da)

1.2.E-09

(b) Feed
1.0.E-09
MF permeate
Detector response (arbitrary units)

8.0.E-10 UF permeate

6.0.E-10

4.0.E-10

2.0.E-10

0.0.E+00
100,000 10,000 1000 100

–2.0.E-10

Apparent molecular weight (Da)

Fig. 4. (a) SEC-UV and (b) SEC-DOC chromatograms for feed, MF permeate and UF permeate.

aromatic and conjugated character were more showing that proportionally more non-UV-
effectively removed by UF than MF (23% and absorbing molecules were removed by UF. Thus
17%, respectively), and the SUVA values both the MF and UF membranes allowed passage
of the lower AMW molecules which, according to
Table 3 Levine et al. [22], in biologically treated sewage
UV Absorbance (UVA) at 254 nm for feed water, MF effluent comprise mainly lower MW humic sub-
and UF permeates stances and proteins, chlorophyll and vitamins.
Although there was little change for UV-
Feed MF UF
absorbing molecules in the 40–70 kDa AMW
water permeate permeate
range for MF permeate (Fig. 4a), there was a
UVA (cm1) 0.345 0.285 0.265 marked reduction in DOC detector response
SUVA (L m1 mg1) 2.78 2.36 2.98 in this range (Fig. 4b), indicating that non-UV-
DOC (mg L1) 12.4 12.1 8.9 absorbing molecules in this size range
214 T. Nguyen et al. / Desalination 236 (2009) 208–215
110
100
90
Transmittance
80
70
60
50
40
500 750 1000 1250
Fig. 5. ATR-FTIR spectra of the foulant layer on MF and UF membranes.
contributed to fouling of the MF membrane. The
organic materials retained by the membranes
in the 40–70 kDa range would be macromole-
cules such as polysaccharides, higher MW humic
acids and proteins, nucleic acids and lipids [22].
For UF, it is clear that molecules in the 40–
70 kDa AMW range, a large proportion of
which were UV-absorbing, contributed a signif-
icant proportion of membrane fouling.
3.5. Characterisation of the foulant layer by
ATR-FTIR
ATR-FTIR analysis of fouled MF and UF
membranes showed similar spectra for all
samples (Fig. 5). The peaks at wave numbers
1640, 1540 and 1040 cm1 indicate that proteins,
polysaccharides and/or aminosugars were
possible foulants [23]. The peak near 1000–
1120 cm1, associated with alcoholic C–O
absorption, and a broad OH band around
3400 cm1, indicate the presence of polysac-
charide-like substances within the fouling layer.
N–H peaks at 3300 and 1550 cm1, together
with a peak at 1640 cm1 corresponding to a
stretching vibration of a C¼O connected to
primary and secondary amides, indicate the
Clean MF
MF of HORS
Clean UF
UF of HORS
1500 1750 2000 2250 2500 2750 3000 3250 3500 3750 4000
Frequency (cm–1)
presence of peptide groups as found in proteins.
Aliphatic CH2 adsorption bands are present at
2853 cm1, and a peak at 1720 cm1 is asso-
ciated with carboxylic groups such as humic and
fulvic acids. Overall, the major compounds
detected in the fouling layers on both MF and
UF membranes were proteins and polysacchar-
ides. This finding was consistent with that pre-
viously reported [24] and supported by several
studies by other researchers [25–27].
4. Conclusions
Comparison of MF and UF of AS-lagoon
effluent from Melbourne’s Western Treatment
Plant showed that MF is more attractive in terms
of greater water recovery. However, as it
retained only a small proportion of the dissolved
organic matter (approximately 2%) which con-
sisted of mostly non-UV-absorbing molecules
of AMW 40–70 kDa, its value as a pre-treatment
for RO is limited.
UF removed approximately 28% of the dis-
solved organic matter, most of which was in the
40–70 kDa size range and a high proportion of
which were UV-absorbing. UF also removed
some organic compounds in the 3–8 kDa range,
 T. Nguyen et al. / Desalination 236 (2009) 208–215
some of which were UV-absorbing. Rapid frac-
tionation demonstrated that the organic matter
removed by UF contained significant propor-
tions of hydrophobic (eg., humic substances),
hydrophilic (eg., polysaccharides) and transphi-
lic fractions, in that order. This was consistent
with EEM data which showed that some
humic-like materials, soluble microbial products
and protein-like extracellular materials were
retained by the UF membrane, whilst fulvic-like
matter largely passed through the membrane. In
accord with these results, ATR-FTIR showed
that the major compounds in the fouling layers
on both MF and UF membranes were primarily
proteinaceous and polysaccharide-like in
character.
These results indicate that UF is preferable to
MF as a pre-treatment for RO, and that further
treatment of the AS-lagoon water is required to
reduce the residual dissolved organic matter
content before a successful RO process can be
implemented.
Acknowledgement
This project was supported by the Smart Water
Fund of Victoria (www.smartwater.com.au).
References
[1] Z. Amjad (Ed.), Reverse Osmosis, Membrane
Technology, Water Chemistry and Industrial Appli-
cation, Van Nostrand Reinhold, New York, 1992.
[2] M. Nystrom, K. Ruohomaki and L. Kaipa, Desali-
nation, 106 (1996) 78–86.
[3] A. Seidel and M. Elimelech, J. Membr. Sci., 203
(2002) 245–255.
[4] S.K. Hong and M. Elimelech, J. Membr. Sci., 132
(1997) 159–181.
[5] A.W. Zularisam, A.F. Ismail and R. Salim, Desa-
lination, 194 (2006) 211–231.
[6] J.W. Cho, G. Amy and J. Pellegrino, Desalination,
127 (2000) 283–289.
215
[7] W. Wei and A.L. Zydney, Environ. Sci. Technol.,
34 (2000) 5043–5050.
[8] L. Fan, J. Harris, F. Roddick and N. Booker, Water
Sci. Technol., Water Suppl., 2 (2002) 313–320.
[9] W. Wei and A.L. Zydney, Desalination, 122
(1999) 63–76.
[10] M. Kabsch-Korbutowicz, K. Majewska-Nowak and
T. Winnicki, Desalination, 126 (1999) 179–185.
[11] W.G. Characklis, In: W.G. Characklis and
K.C. Marshall (Eds.), Biofilms, Wiley, New York,
1990.
[12] T.F. Speth, R.S.Summers and A.M. Gusses,
Environ. Sci. Technol., 32 (1998) 3612–3617.
[13] K. Kimura, Y. Hane, Y. Watanabe, G. Amy and
N. Ohkuma, Water Res., 38 (2004) 3431–3441.
[14] I.C. Escobar, E. Hoek, C.J. Gabelich, F. di Giano,
et al., J. AWWA, 97 (8) (2005) 79–89.
[15] C. Gabelich, T. Yun, B. Coffey and I.H. Suffet,
Desalination, 154 (2003) 207–223.
[16] S.G. Yiantsios and A.J. Karabelas, Desalination,
118 (1998) 143–152.
[17] S.L. Kim, J.P. Chen, Y.P. Ting, Sep. Purif. Tech-
nol., 29 (2002) 171–179.
[18] G. Caper, L. Yilmaz and U. Yetis J. Hazard.
Mater. (2007) (Article in press), DOI: 10.1016/
j.jhazmat.2007.06.100.
[19] C.V. Vedavyasan, Desalination, 132 (2000)
345–347.
[20] G.-P. Sheng and H.-Q. Yu, Water Res., 40 (2006)
1233–1239.
[21] M.L. Nguyen, P. Westerhoff, L. Baker, Q. Hu,
M. Esparza-Soto and M. Sommerfeld, J. Environ.
Eng., 131 (2005) 1574–1582.
[22] A.D. Levine, G. Tchobanoglous and T.J. Asano,
Water Pollut. Control Fed., 57 (7) (1985)
805–816.
[23] S.E. Barrett, S.W. Krasner and G.L. Amy (Eds),
ACS Symposum Series 761, American Chemical
Society, Washington DC, 2000.
[24] D. Stork, T. Nguyen, F.A. Roddick and J.L. Harris,
Proceedings AWA Membrane Specialty Confer-
ence II, 21–23 Feb. 2007, CD-ROM, Melbourne,
Australia.
[25] C. Jarusutthirak, G. Amy and J.-P. Croué, Desali-
nation, 145 (2002) 247–255.
[26] N. Her, G. Amy, H.-R. Park, M. Song, Water Res.,
38 (2004) 1427–1438.
[27] H. K. Shon and S. Vigneswaran, In S. Kim, J. Cho
and H.H. Ngo, J. Membr. Sci., 278 (2006) 232–238.