This document evaluates the performance of a spiral-wound ultrafiltration pilot plant for direct drinking water treatment over a two-year period. Results showed that using periodic relaxation with flushing and chlorine dosage increased permeate production by 49% compared to no such procedures. Continuous operation demonstrated better performance in fall and winter, associated with fewer algae blooms. Clarifying and recirculating concentrate allowed reaching 99% water recovery. Water quality monitoring found the system effectively removed pathogens, turbidity and color.
2. performance [16–20]. Evaluated approaches to improve membrane
systems' performance include feed pre-treatment, addition of chemicals
combined with activated carbon and oxidizers, membrane backwashing,
and chemical cleaning, among others [18,21–25].
Considering the limited use of spiral-wound ultrafiltration sys-
tems for drinking water treatment, fouling issues on these systems,
and consequently their performance, are mostly addressed by feed
water pre-treatment, addition of chemicals, and membrane chemical
cleaning. However, recent studies have shown that combining tradition-
al procedures with specific operational conditions can result in mem-
brane performance improvement [26]. Considering the results obtained
in previous studies, it was deemed necessary to produce a better evalua-
tion of the influence of pre-chlorination and periodic relaxation and
flushing procedure on the operational performance of a spiral-wound ul-
trafiltration membrane for direct drinking water treatment, using a pilot
plant installed at the Guarapiranga's reservoir, located in the Sao Paulo
Metropolitan Region. Evaluation tests last 2 years, which made possible
an evaluation of the influence of specific operational procedures and sea-
sonal weather conditions on system performance.
2. Experimental
2.1. Pilot plant unit
All the experiments had been developed using an ultrafiltration pilot
plant installed close to the Guarapiranga Reservoir, in the city of Sao
Paulo, in an area that belongs to the Companhia de Saneamento
Basicodo Estado de Sao Paulo (SABESP), the company responsible for
drinking water treatment and distribution in Sao Paulo State — Brazil.
This reservoir is used to produce drinking water for a population of ap-
proximately 6 million people.
A flow diagram of the ultrafiltration pilot plant used in the experi-
ments is depicted in Fig. 1, where bold lines represent the main flow
and dashed lines represent the flow during chemical cleaning, or filters
backwash. Only one spiral-wound ultrafiltration membrane from
GE-Osmonics (Model PW-4040F) was used, with a molecular weight
cut-off of 10,000 g·mol−1
. The pilot plant operation was fully automat-
ed, and once the operational conditions were set, it only stopped by op-
erator interference, or by a low water level in the feed tank.
Water from the reservoir was diverted from a main water transfer
pipeline from Guarapiranga Reservoir to the Alto da Boa Vista drinking
water treatment plant. Before reaching the feed tank, reservoir water
passed through a sand filter with a 0.5 mm effective diameter of sand
particles and a filtration area of 0.19 m2
(19CFA4-M — Jacuzzi do Brasil).
From the feed tank, water was pumped to the membrane pressure ves-
sel, passing through a screen filter of 100 μm (1″ Super — Amiad Water
Systems). From the membrane pressure vessel two streams were
obtained, including permeate which was sent to the cleaning tank and
then to the permeate tank, and concentrate which was recirculated to
the feed pump suction line, in order to increase water recovery. Mem-
brane operation pressure was set using a globe valve (G1). Ultrafiltration
pilot plant was operated in the feed and bleed mode [22], in order to
facilitate its operation when working with an increased water recovery,
because of problems for keeping continuous concentrate discharge
observed in a previous study [26]. For controlling water recovery, a set
of timer-operated solenoid valves was used. The timer (K1) was
programmed to activate solenoid valves S2 and S3 for 10 s every
10 min, for concentrate discharge.
Chemical cleaning was performed periodically, using first permeate
water for flushing and rinsing, then sodium hydroxide solution (pH 12),
for organic fouling removal, and finally peracetic acid solution (0.1%) for
sanitization. Membrane cleaning procedure was used either when per-
meate flow declined, which resulted in a variable cleaning frequency, or
after a specific test.
The system's operational performance was evaluated through con-
tinuous monitoring of permeate flow (FE-1), temperature (TE) and
turbidity (AE — from Hach — model 1720E), recirculation flow (FE-2),
membrane pressure (PE) and head loss (dPE). These data were
Feed Tank
Permeate
Tank
Feed
pump
Cleaning pump
Cleaning tank
Chlorine Tank Dosing
pump
Reservior
S-4 (NC)
Screen
filter
Sand Filter
Concentrate
S-1 (NC)
S-2 (NC)
Pressure Vessel
Permeate
Drainage
dPE
PE
P-15
FE1 TE
AE
G-1
S-3 (NO)
Data Logger
K1
K2FE2
FE – Flow transducer
PE – Pressure transducer
TE – Temperature transducer
dPE – Differential pressure transducer
AE – Turbidity transducer
K – Timer
G – Globe valve
S – Solenoid valve
NO – Normally open
NC – Normally closed
– Ball valve
– Check valve
Fig. 1. Flow diagram of the ultrafiltration pilot plant used in the evaluation.
69J.C. Mierzwa et al. / Desalination 307 (2012) 68–75
3. collected and stored in the data logger every 2 min, and once a week, or
after a specific test, stored data were downloaded to a computer for fur-
ther analysis.
2.2. Operational procedures tests
In order to establish specific operational conditions, preliminary
tests were carried out to evaluate the influence of chlorine dosage
in the feed tank and the use of periodic relaxation procedure on
membrane performance. For this purpose three operational condi-
tions were evaluated: (a) periodic relaxation and flushing procedure
without chlorine dosage, (b) periodic relaxation and fluxing proce-
dure with chlorine dosage, and (c) no periodic relaxation and flushing
procedure with chlorine dosage.
The periodic relaxation and flushing procedure consisted of a daily
interruption of system operation controlled by another timer (K2).
After a continuous 24-hour operation period, the system was turned
off for 10 min. During the stopping period, after 4 min the same timer
(K2) activated the feed pump and solenoids S1 and S3 to perform mem-
brane flushing for 2 min, turning the system off again, for 4 more mi-
nutes, and after that normal system operation was resumed.
2.3. System performance evaluation
Once the best operation procedure was established the pilot plant
stated to operate continuously for a long-term performance evaluation,
from December 2007 to December 2009, resulting in a continuous oper-
ation time of almost 16,000 h. During this period, permeate flow, temper-
ature and turbidity, membrane pressure and head loss, and concentrate
recirculation flow measurements were collected every 2 min and stored
in the data logger. Data in the data logger were downloaded to a comput-
er once a week, for treatment and analysis.
For the evaluation of water treatment efficiency, raw water and per-
meate samples were collected periodically to evaluate UV-254 absorp-
tion (UV-Mini 1240 — Shimadzu), dissolved organic carbon (TOC-V
CPH — Shimadzu Corporation), apparentcolor (AcquaColor — Policontrol
Instrumentos Analíticos), pH (Q400MT — Quimis Aparelhos Científicos),
total coliforms and Escherichia coli, using the plate count method
(Colilert® — IDEXX Laboratories).
2.4. Concentrate treatment and recirculation
In order to increase system water recovery and solve the problem
with concentrate disposal, a test for concentrate clarification and
recirculation to the feed tank was performed. Preliminary concentrate
clarification tests were carried out using a Jar-test equipment (JT-203/
06 — Milan Equipamentos Científicos Ltda), to determine the best clar-
ification condition. Ferric chloride hexahydrate was used as a coagulant,
with a ferric ion concentration ranging from 2 to 17 mg Fe3+
·L−1
. The
pH ranged from 4.5 to 7.0 using analytical grade sodium hydroxide or
hydrochloric acid solutions prepared using analytical grade reagents.
Clarification efficiency was evaluated based on apparent color reduction.
0
50
100
150
200
250
300
PermeateFlux(L.h-1.m-2)
Operation time (minutes)
0 1000 2000 3000 4000
Operation time (minutes)
0 1000 2000 3000 4000
Operation time (minutes)
0 1000 2000 3000 4000
b
a) b) c)
Fig. 3. Results from operational procedure tests: a) continuous operation with chlorine dosage; b) periodic relaxation and flushing with chlorine dosage; and c) periodic relaxation
and flushing without chlorine.
Clarifier
Ultrafiltration Feed Tank
Floculation Tank
M
Concentrate
Storage Tank
Pump
Dosing pump
Ferric Chloride
Tank
Concentrate
UF Feed pump
Sewer
Fig. 2. Concentrate clarification system.
70 J.C. Mierzwa et al. / Desalination 307 (2012) 68–75
4. Jar-test operational conditions used for obtaining the coagulation dia-
gram were:
• rapid mixing: 120 s−1
for 2 min;
• slow mixing: 40 s−1
for 20 min;
• sedimentation time: 20 min.
Samples for analysis were drawn 7.5 cm from the surface of each
jar, which resulted in a settling velocity of 6.25×10−3
cm·s−1
.
With the obtained optimal clarification conditions, a continuous
clarification unit (Fig. 2) with a capacity of 20 L·h−1
, which consisted
of an accumulation tank (70 L), a stirred flocculation tank (5 L), and a
settling tank (70 L), was installed close to the ultrafiltration pilot
plant, and the system started to operate with full clarified concentrate
recirculation. The sludge produced was discharged in the sewer, be-
cause of the small amount produced.
3. Results and discussion
3.1. Operational procedures tests
Results from the three operational tests, considering periodic relax-
ation and fluxing procedure with and without chlorine dosage and con-
tinuous operation with chorine dosage are presented in Fig. 3. The
chlorine dosage used in the tests was 2.0 mg·L−1
as active Cl2.
As it can be observed in Fig. 3, the best system performance, consid-
ering the permeate production, was obtained with the combination of
periodic relaxation and flushing procedure and chlorine dosage. Consid-
ering the same period of operation, almost 67 h, for the three operation-
al conditions, the greatest volume of permeate produced was 49%
higher than that produced when no periodic relaxation and flushing
procedure was used. Results obtained in the three performance tests
are presented in Table 1. It is clear that only the combination of periodic
relaxation and flushing procedure with chlorine dosage resulted in a
significant increase of system performance. During the relaxation pro-
cedure, contaminants that had been transported to the membrane
surface by convection, mainly colloids, became loose, and when the
flushing started most of these contaminants were removed from the
membrane surface, which reduced permeate flow resistance for the
new production cycle. Once contaminants had been removed from the
membrane surface, chlorine effectiveness improved, causing a delay in
the biofilm establishment, which contributed to an enhancement in
overall system performance. The effect of relaxation and flushing proce-
dure on membrane resistance can clearly be observed in Fig. 3.b, and the
influence of chlorine dosage on performance improvement when
Fig. 3.c and Fig. 3.b are compared.
3.2. System performance evaluation
Once the best operational procedure was established, the long term
ultrafiltration pilot plant performance evaluation began. System opera-
tional conditions obtained during the evaluation period, 2 years, are
presented in Table 2.
Data presented in Table 2 shows that the system operated with a rel-
ative low pressure, with an average flux close to the one specified by the
membrane supplier. According to the PW4040F membrane specifica-
tion sheet, pure-water flux for this membrane ranges from 15 to
40 L·h−1
·m−2
with a typical pressure ranging from 555 to 931 kPa
00
50
100
150
200
250
300
350
400
450
500
18/12/2007a04/01/2008
04/01a14/01/2008
14/01a16/01/2008
16/01a28/01/2008
29/01a07/02/2008
07/02a08/02/2008
12//02a19/02/2008
19/02a21/02/2008
26/02a06/03/2008
06/03a18/03/2008
18/03a27/03/2008
28/03a03/04/2008
03/04a08/04/2008
08/04a24/04/2008
25/04a07/05/2008
08/05a21/05/2008
21/05a07/06/2008
07/06a16/06/2008
16/06a10/07/2008
11/07a24/07/2008
24/07a07/08/2008
07/08a19/08/2008
19/08a02/09/2008
05/09a16/09/2008
16/09a22/09/2008
22/09a02/10/2008
02/10a07/10/2008
08/10a14/10/2008
14/10a22/10/2008
23/10a29/10/2008
29/10a03/11/2008
03/11a19/11/2008
21/11a04/12/2008
05/12a22/12/2008
23/12/2008a06/01/2009
07/01a16/01/2009
16/01a22/01/2009
26/01a05/02/2009
05/02a09/02/2009
19/03a01/04/2009
01/04a17/04/2009
17/04a29/04/2009
05/05a21/05/2009
21/05a04/06/2009
05/06a12/06/2009
12/06a18/06/2009
18/06a24/06/2009
24/06a03/07/2009
03/07a08/07/2009
08/07a14/07/2009
14/07a28/07/2009
28/07a06/08/2009
06/08a12/08/2009
12/08a19/08/2009
19/08a27/08/2009
27/08a04/09/2009
04/09a10/09/2009
11/09a16/09/2009
16/09a02/10/2009
02/10a19/10/2009
19/10a29/10/2009
29/10a06/11/2009
10/11a18/11/2009
18/11a27/11/2009
01/12a09/12/2009
09/12a18/12/2009
18/12a23/12/2009
Summer Fall Winter Spring Summer Fall Winter Spring
NormalizedPermeateFlow(L.h-1)
Operation Period
Minimum Average Maximum 2 per. Mov. Avg. (Average)
Fig. 4. Observed normalized permeate flow according to annual seasons.
Table 2
Ultrafiltration pilot plant operational conditions from December 2007 to December
2009.
Operational parameter Values
Minimum Average Maximum
Transmembrane pressure (kPa) –x– 106.3 –x–
Permeate flux (L·h−1
·m−2
) 10.6 15.5 31.1
Recirculation flow (m3
·h−1
) 0.8 2.3 2.7
Water recovery (%) 83.6 88.5 93.4
Permeate Turbidity (NTU) 0.03 0.04 0.6
Table 1
Operation time, average permeate flow and produced permeate volume obtained during
system performance evaluation tests.
Operational procedure Operation
time (hours)
Average permeate
flux (L·h−1
·m−2
)
Permeate
volume (m3
)
Continuous operation with
chlorine dosage
67.3 167.4 11.3
Periodic relaxation and
flushing without
chlorine dosage
191.2 12.9
Periodic relaxation and
flushing with chlorine
dosage
249.9 16.8
71J.C. Mierzwa et al. / Desalination 307 (2012) 68–75
5. [27]. It is also important to observe that water recovery was very
high, close to 90%, and the permeate turbidity was kept well below
the Brazilian standard for drinking water, which is 1.0 NTU. The av-
erage raw water turbidity was 3.45 NTU.
The behavior of system normalized permeate flow during the evalu-
ation period can be observed in Fig. 4, which shows a permeate flow
improvement during the fall and winter seasons. This behavior can be
attributed to the lower temperatures observed in these periods when
compared to spring and summer, which directly affects Guarapiranga
Reservoir's water quality, probably because of algae blooms. This state-
ment can be supported by the results for chlorophyll-a analysis
performed by the São Paulo Environmental Agency (CETESB), which
are presented in Table 3.
Comparing the results presented in Table 3, and the specific period of
operation depicted in Fig. 4, it is possible to establish a direct correlation
among system normalized permeate flow, chlorophyll-a concentration,
and water temperature. This implies that when chlorophyll-a concentra-
tion and water temperature decrease, during fall and winter seasons,
permeate flow increases, then starts to decrease in the spring season,
reaching its lowest values in summer. This behavior becomes more evi-
dent through the analysis of data presented in Table 4.
Through statistical analysis, using the ANOVA-single factor with a
confidence level of 95%, it could be verified that average values for the
permeate flow data set of 2008 and 2009 belong to the same popula-
tion with a probability of 29.7% (p-value of 0.297), which supports
the correlation between system performance and seasonal weather
conditions.
The influence of seasonal weather on the operational performance
of the ultrafiltration pilot unit can also be demonstrated based on the
frequency of chemical cleanings in each season, as it is presented in
Fig. 5. It is important to note that the frequency of chemical cleanings
is higher in spring and summer, which supports previous statements
about the influence of seasons on membrane operational performance.
With regard to water treatment efficiency, Table 5 presents results
for raw and permeate water samples collected and analyzed during
the ultrafiltration pilot unit operation, and Fig. 6 presents removal
efficiencies.
Results presented in Table 5 and Fig. 6 clearly demonstrate the high
performance of the ultrafiltration pilot unit for direct drinking water
treatment, mainly because the high removal efficiencies obtained for
potential pathogenic organisms, 100% for E. coli and total coliforms,
95.1% for turbidity, and 91.5 for apparent color. Even for dissolved or-
ganic matter, indirectly measured through DOC and UV-254 absorption,
a significant removal was obtained, with average values of 28.7% and
41.7%, respectively. It is important to mention that only one permeate
sample presented positive results for total coliform, very close to the
method detection limit. This can most probably be associated to sample
contamination during its manipulation.
Results for UV-254 removal obtained in this study (41.7%) are very
similar to those obtained by Liang et al. [8], which were only obtained
because of chemical additions and pretreatment used on their exper-
iments. However permeate fluxes obtained in the experiments devel-
oped by the former researchers (60 to 70 L·h−1
·m−2
), were higher
than those obtained in this study (15.5 L·h−1
·m−2
), under the
same operational pressure, which could be attributed to the fact
that they used a membrane with a higher MWCO (100,000 g·mol−1
).
The results thus far presented clearly demonstrate the high poten-
tial for using spiral-wound ultrafiltration membranes for drinking
water treatment, considering stable operational conditions and high
efficiencies for contaminant removals, and that specific operational
procedures can improve long-term system performance.
3.3. Concentrate treatment and recirculation
Considering the problems associated with the concentrate produced
when membrane technology is applied for water treatment [30], since
soluble salt concentration is not an issue, ultrafiltration concentrate
treatment and recirculation can be a good option to cope with this
by-product. A coagulation diagram obtained in laboratory clarification
tests is presented in Fig. 7, and average concentrate characteristics are
presented in Table 6.
Based on the coagulation diagram obtained (Fig. 7), the best condi-
tions for concentrate clarification are a pH lower than 6.0 and a ferric
ion dosage higher than 12 mg Fe3+
·L−1
, which result in a color remov-
al efficiency above 80%. These results can be explained based on concen-
trate characteristics, high color and low turbidity (Table 6), indicating
that natural organic matter (NOM) is the main contaminant to be re-
moved, which requires a lower coagulation pH in order to obtain good
removal efficiencies [31].
Once the coagulation and flocculation conditions were determined,
the clarification unit started to operate in order to evaluate the influ-
ence of clarified concentrate recirculation on the ultrafiltration pilot
0.0
0.5
1.0
1.5
2.0
ChemicalCleaning
Frequency(month-1)Sum
m
er/2008
Fall/2008W
inter/2008Spring/2008Sum
m
er/2009
Fall/2009W
inter/2009Spring/2009
Fig. 5. Frequency of chemical cleanings according to weather seasons.
Table 4
Seasonal variation of permeate flow of the ultrafiltration pilot plant in 2008 and 2009.
Year Season Permeate flow (L·h−1
)
Minimum Average Maximum
2008 Summer 81.7 116.3 155
Fall 95.5 117.9 240.4
Winter 101.6 139.8 332.8
Spring 103.7 134.9 305.3
2009 Summer 43.3 86.3 191.6
Fall 62.7 124.8 263.7
Winter 89.9 130.6 266.5
Spring 78.6 114.9 216.9
Table 3
Results for chlorophyll-a concentrations measured at the Guarapiranga Reservoir during 2008 and 2009 [28,29].
Year Chlorophyll-a (μg·L−1
) and water temperature (°C)
January March May July September November
2008 20.5 23.5 n.a. 27.3 38.6 21.5 14.9 19.3 26 19.5 53.5 23.1
2009 7.7 23.5 9.2 29.1 6 21.7 15.1 17.8 33.2 21.1 71.2 25.2
n.a. — not analyzed.
72 J.C. Mierzwa et al. / Desalination 307 (2012) 68–75
6. unit performance. Coagulant dosage was 12.4 mg Fe3+
·L−1
, and no
acid or alkali was added because the coagulant dosage was sufficient
for pH adjustment. Table 7 presents the results for the removal of spe-
cific contaminants, obtained during the clarification unit operation.
As can be seen in Table 7, the clarification system was very efficient
in the removal of apparent color (94.9%), and turbidity (98.7%), but with
significant removal efficiencies for UV-254 light absorption (69.8%) and
DOC (61.5%). These results are in agreement with the data presented in
the literature [31]. The quality of the clarification system effluent
(Table 7) is quite better than the quality of the raw ultrafiltration pilot
unit feed (Table 5), which indicates the feasibility of clarified concen-
trate recirculation.
The influence of clarified concentrate recirculation on the ultrafil-
tration pilot unit performance was evaluated by comparing permeate
flow and online turbidity, before and after the recirculation procedure
was started. The results are presented in Fig. 8.
Due to operational difficulties, the clarification unit only operated
for 54 h with continuous clarified recirculation to the ultrafiltration
system feed tank. The main difficulty faced during the experiment
was to keep the clarification system working in a steady state condi-
tion, because most of the system components were adapted for the
intended use, and operation flow (20 L·h−1
), was higher than con-
centrate flow production, averaging 14 L·h−1
. Although the concen-
trate clarification system did not operate continuously for a long
period of time, by the analysis of Fig. 8 it is possible to observe a
small change in the operational performance of ultrafiltration pilot
unit. From the start of the clarified concentrate recirculation, no sig-
nificant change in permeate flow was observed, but permeate turbid-
ity became more stable. However, it was possible to notice a steady
increase in transmembrane pressure until the periodic relaxation
and flushing procedure started.
A similar behavior described previously occurred in the subse-
quent operation cycle, but with a noticeable reduction in permeate
flow. Once the concentrate recirculation was stopped, transmem-
brane pressure stabilized, permeate flow started to decrease steadily,
and permeate turbidity became more unstable. Since concentrate
recirculation represented less than 15% of the ultrafiltration pilot
unit feed flow, and the clarified concentrate characteristic seems to
be quite similar to raw water, it is not possible to draw a definitive
conclusion about the observed behavior. However, since clarified con-
centrate was not filtered, and no pH adjustment was made before
recirculation, it is possible that residual ferric ions were precipitated
and flocculated. This flocculated material was then retained on the
membrane surface, increasing transmembrane pressure during nor-
mal system operation, and during the periodic relaxation and flushing
procedure this material was removed, resulting in a reduction in the
transmembrane pressure at the beginning of the next production
cycle. Observed turbidity peaks in Fig. 8 are most probably related
to bubbles or other hydraulic phenomena inside the turbidimeter
cell, because these peaks were observed for a short period of time,
mainly after the relaxation and flushing procedure was finished.
Even though it was not possible to draw a definitive conclusion about
the influence of clarified concentrate recirculation on the ultrafiltration
4.5
5.0
5.5
6.0
6.5
7.0
2 4 6 8 12 17
pH(unitis)
Fe3+ (mg.L-1)
0%-20% 20%-40% 40%-60% 60%-80% 80%-100%
Fig. 7. Ultrafiltration concentrate coagulation diagram.
0
20
40
60
80
100
ABS UV-
254
DOC Aparent
Color
Turbidity E. Coli Total
Coliforms
Removal(%)
Fig. 6. Contaminant removal efficiencies obtained by the ultrafiltration pilot unit during
the evaluation period.
Table 5
Results for raw and permeate water samples collected during the evaluation period.
Sample ABS UV-254 (cm−1
) DOC (mg·L−1
) Aparent color (uC) Turbidity (NTU)a
E. coli (NMP·100 mL−1
) Total coliform (NMP·100 mL−1
)
Raw water Minimum 0.03 2.9 25.3 1 0 63
Average 0.1 4.8 61.4 3.5 49 824
Maximum 0.31 12.9 190 11.2 306 2420
Standard deviation 0.05 2.2 39 3 81 974
Permeate Minimum 0.02 1.9 2 0.1 0 0
Average 0.06 3.4 5 0.2 0 0
Maximum 0.08 11.6 13.3 0.8 0 2
Standard deviation 0.02 1.9 2.8 0.2 0 1
Number of samples 51 47 20 23 16 14
uC — color unit (mg Pt-Co·L−1
).
(AP 2000 — Policontrol Instrumentos Analíticos).
a
Presented turbidity results were obtained using a portable turbidimeter.
73J.C. Mierzwa et al. / Desalination 307 (2012) 68–75
7. performance, this procedure can increase water system recovery up to
99%, considering only a small amount will be lost with the produced
sludge. This figure, considering the problems related to water scarcity
faced in the Metropolitan Region of Sao Paulo, and many other regions
in Brazil, makes the concentrate clarification and recirculation scheme
a suitable alternative to be better explored. This means that more efforts
should be applied in order to better understand the impact of ultrafiltra-
tion concentrate recirculation on the system operational performance,
through long term experiments.
4. Conclusions
Based on the results obtained in the evaluation of spiral-wound
ultrafiltration membranes for drinking water treatment, as presented
in this study, it is possible to conclude that:
1- Different operational procedures can expressively affect spiral-wound
ultrafiltration performance for water treatment. Periodic relaxation
and flushing procedure, along with chlorine addition, was able to in-
crease permeate production by almost 50%. Since the frequency of re-
laxation and flushing procedure was not changed during long-term
performance evaluation, complementary studies are required to ver-
ify if higher frequencies can result in better performance results.
2- A long-term operational performance evaluation clearly demon-
strated the influence of seasons on ultrafiltration operational perfor-
mance. Higher permeate production and lower chemical cleaning
frequencies were obtained during fall and winter seasons. This be-
havior, for a humid subtropical climate, can be explained by the
higher primary productivity in the reservoir during spring and sum-
mer seasons, resulting in algae blooms during these periods.
3- Ultrafiltration concentrate clarification using ferric chloride was
quite effective in removing NOM, indirectly evaluated through
DOC and UV-254 absorption measurements, and turbidity, allowing
its recirculation to the ultrafiltration feed tank. This procedure, even
considering the short evaluation period, made it possible to increase
the overall system water recovery to almost 99%, and at the same
time it addressed one of the major issues about membrane technol-
ogy for water treatment, which is the final concentrate disposal.
Even with these promising results, additional clarification and
recirculation tests should be carried out in order to drawn a more re-
liable conclusions.
4- Finally, long term ultrafiltration pilot plant operation demonstrat-
ed its performance consistency on producing treated water with
very good quality, mainly related to the removal of potential path-
ogenic organisms, NOM, and turbidity, indicating the process fea-
sibility for direct drinking water treatment in the SPMR.
Table 7
Results for specific contaminant removal efficiencies obtained in the clarification unit.
Sample ABS UV-254
(cm−1
)
DOC
(mg·L−1
)
Aparent
color (uC)
Turbidity
(NTU)
Feed Minimum 0.17 11 64 3.1
Average 0.28 12.5 117 9.8
Maximum 0.43 13.7 204 22.2
Standard deviation 0.11 1.2 64 9
Clarified Minimum 0.04 3.5 2 0
Average 0.09 4.8 6 0.1
Maximum 0.13 6.5 20 0.5
Standard deviation 0.03 1 5 0.2
Number of samples 18 19 19 18
Average removal efficiency (%) 69.8 61.5 94.9 98.7
uC — color unit (mg Pt-Co·L−1
).
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0
50
100
150
200
250
0 5,000 10,000 15,000 20,000
PermeateTurbidity(NTU)
NormalizedPermeateFlow(L.h-1)and
TransmembranePressure(kPa)
Transmembrane Pressure (kPa) Normalized Permeate Flow (L/h) Turbidity (NTU)
Permeate
Flow
Transmembrane
Pressure
Turbidity
Continuos Clarified
Concentrate Recirculation
Stop
Continuous Clarified
Concentrate Recirculation
Start
ration time (minutes)
Fig. 8. Comparative ultrafiltration pilot unit operational performance before and after clarified concentrate recirculation.
Table 6
Ultrafiltration pilot unit concentrate characteristics.
Quality variable Number of samples Unit Minimum Average Maximum Standard deviation
ABS UV-254 51 cm−1
0.09 0.23 0.49 0.07
Dissolved organic carbon (DOC) 47 mg·L−1
4.1 10.98 24.60 4.09
Aparent color 20 uC 103 169 358 67
E. coli 16 NMP·100 mL−1
0 5 20 7
Total coliforms 14 NMP·100 mL−1
0 327 2420 659
Turbidity 23 NTU 2.6 14.3 45.8 12.9
uC — color unit (mg Pt-Co·L−1
).
74 J.C. Mierzwa et al. / Desalination 307 (2012) 68–75
8. Acknowledgments
The authors of this paper want to express their gratitude to the
Finaciadora de Estudos e Projeto (FINEP), for their financial support,
and to the Companhia de Saneamento Basico do Estado de Sao
Paulo (SABESP), for allowing the installation and operation of the ul-
trafiltration pilot unit at the Guarapiranga Reservoir. A special thanks
to Tyler Clites for the text revision.
References
[1] J.G. Jacangelo, R.R. Trussell, M. Watson, Role of membrane technology in drinking
water treatment in the United States, Desalination 113 (1997) 119–127.
[2] S.S. Madaeni, The application of membrane technology for water disinfection,
Water Resour. 33 (2) (1999) 301–308.
[3] J.M. Arnal, B. Garcia-Fayos, G. Verdu, J. Lora, Ultrafiltration as an alternative mem-
brane technology to obtain safe drinking water from surface water: 10 years of
experience on the scope of the AQUAPOT project, Desalination 248 (2009) 34–41.
[4] W. Ma, Z. Sun, Z. Wang, Y.B. Feng, T.C. Wang, U.S. Chart, C.H. Miu, S. Zhu, Application
of membrane technology for drinking water, Desalination 119 (1998) 127–131.
[5] B. Nicolaisen, Developments in membrane technology for water treatment, Desa-
lination 153 (2002) 355–360.
[6] J.M. Arnal, M. Sancho, G. Verdú, J. Lora, J.F. Marin, J. Chiller, Selection of the most
suitable ultrafiltration membrane for water disinfection in developing countries,
Desalination 168 (2004) 265–270.
[7] X. Shengjia, L. Xingb, Y. Jic, D. Bingzhia, Y. Juanjuana, Application of membrane tech-
niques to produce drinking water in China, Desalination 222 (2008) 497–501.
[8] H. Liang, W. Gongb, G. Li, Performance evaluation of water treatment ultrafiltra-
tion pilot plants treating algae-rich reservoir water, Desalination 221 (2008)
345–350.
[9] J.C. Rojas, B. Moreno, G. Garralón, F. Plaza, J. Pérez, M.A. Gómez, Potabilization of
low NOM reservoir water by ultrafiltration spiral wound membranes, J. Hazard.
Mater. 158 (2008) 593–598.
[10] J.M. Arnal, B. García-Fayos, M. Sancho, G. Verdú, J. Lora, Design and installation of
a decentralized drinking water system based on ultrafiltration in Mozambique,
Desalination 250 (2010) 613–617.
[11] The Freedonia Group, Executive summary, world membrane separation technologies —
industry study with forecasts for 2012 & 2017. Study #2468 , April 2009.
[12] T. Leiknes, Membrane technology in environmental engineering — meeting future
demands and challenges of the water and sanitation sector, Desalination 199
(2006) 12–14.
[13] G. Crozes, C. Anselme, G. Mallevialle, Effect of adsorption of organic matter on
fouling of ultrafiltration membranes, J. Membr. Sci. 84 (1993) 61–77.
[14] A.R. Costa, M.N. de Pinho, M. Elimelech, Mechanisms of colloidal natural organic
matter fouling in ultrafiltration, J. Membr. Sci. 281 (2006) 716–725.
[15] W. Gao, Heng Liang, Jun Ma, Mei Han, Zhong-lin Chen, Zheng-shuang Han,
Gui-bai Li, Membrane fouling control in ultrafiltration technology for drinking
water production: a review, Desalination 272 (2011) 1–8.
[16] A.I. Schäfer, A.G. Fane, T.D. Waite, Fouling effects on rejection in the membrane
filtration of natural waters, Desalination 131 (2000) 215–224.
[17] N. Lee, G. Amya, J.P. Croue, H. Buisson, Identification and understanding of fouling
in low-pressure membrane (MF/UF) filtration by natural organic matter (NOM),
Water Res. 38 (2004) 4511–4523.
[18] Y. Chen, B.Z. Dong, N.Y. Gao, J.C. Fan, Effect of coagulation pretreatment on fouling
of an ultrafiltration membrane, Desalination 204 (2007) 181–188.
[19] D.B. Mosqueda-Jimenez, P.M. Huck, O.D. Basu, Fouling characteristics of an ultra-
filtration membrane used in drinking water treatment, Desalination 230 (2008)
79–91.
[20] W. Neubrand, S. Vogler, M. Ernst, M. Jekel, Lab and pilot scale investigations on
membrane fouling during the ultrafiltration of surface water, Desalination 250
(2010) 968–972.
[21] K. Konieczny, G. Klomfas, Using activated carbon to improve natural water treat-
ment by porous membranes, Desalination 147 (2002) 109–116.
[22] H. Choi, H.S. Kim, I.T. Yeom, D.D. Dionysiou, Pilot plant study of an ultrafiltration
membrane system for drinking water treatment operated in the feed-and-bleed
mode, Desalination 172 (2005) 281–291.
[23] H. Liang, W. Gong, J. Chen, G. Li, Cleaning of fouled ultrafiltration (UF) membrane
by algae during reservoir water treatment, Desalination 220 (2008) 267–272.
[24] G. Goldman, J. Starosvetsky, R. Armon, Inhibition of biofilm formation on UF
membrane by use of specific bacteriophages, J. Membr. Sci. 342 (2009) 145–152.
[25] A.W. Zularisam, A.F. Ismail, M.R. Salim, M. Sakinah, T. Matsuura, Application of
coagulation–ultrafiltration hybrid process for drinking water treatment: optimi-
zation of operating conditions using experimental design, Sep. Purif. Technol.
65 (2009) 193–210.
[26] J.C. Mierzwa, I. Hespanhol, M.C.C. da Silva, L.D.B. Rodrigues, C.F. Giorgi, Direct drink-
ing water treatment by spiral-wound ultrafiltration membranes, Desalination 230
(2008) 41–50.
[27] GE Power & Water — Water & Process Technologies, PW Series Ultrafiltration —
Post treatment of RO and NF, Fact sheet. Available at: http://www.gewater.com/
pdf/Fact%20Sheets_Cust/Americas/English/FS1287EN.pdf 2010.
[28] Companhia de Tecnologia de Saneamento Ambiental (CETESB), Qualidade das águas
interiores do Estado de São Paulo — 2008, 2009. Report available at: http://www.
cetesb.sp.gov.br/agua/aguas-superficiais/35-publicacoes-/-relatorios.
[29] Companhia de Tecnologia de Saneamento Ambiental (CETESB), Qualidade das águas
interiores do Estado de São Paulo — 2009, 2010. Report available at: http://www.
cetesb.sp.gov.br/agua/aguas-superficiais/35-publicacoes-/-relatorios.
[30] M.C. Mickley, Membrane concentrate disposal: practices and regulations. Water
Treatment Engineering and Research Group, U.S. Department of the Interior, Bu-
reau of Reclamation, Denver CO, Report nº 123, second edition, 2006. available at:
http://www.usbr.gov/pmts/water/publications/reportpdfs/report123.pdf.
[31] A. Matilainen, M. Vepsäläinen, M. Sillanpää, Natural organic matter removal by coag-
ulation during drinking water treatment: a review, Adv. Colloid Interface Sci. 159
(2010) 189–197.
75J.C. Mierzwa et al. / Desalination 307 (2012) 68–75