Effect of pH on Fouling Attachments and Power Consumption in

Ultraltration of Latex Solution

Amira Abdelrasoul,* Huu Doan and Ali Lohi
Department of Chemical Engineering, Ryerson University, 350 Victoria Street, Toronto, Ontario, Canada M5B 2K3

The goal of the present study was to remediate ultraltration fouling through the altering of the latex paint solution pH, in order to reduce the total
mass fouling and the specic power consumption. Polycarbonate at membrane with a pore size of 0.05 mm was used under a constant feed ow rate
and crossow mode in ultraltration of a latex paint solution. It was observed that the ionic strength of the latex solution had a signicant effect on the
membrane hydrophilicy and the particle surface charge, which in turn inuenced the particleparticle and particlemembrane attachment. At the
transmembrane pressure of 15 psi, a feed ow rate of 1 LPM, and a feed concentration of 1.3 kg/m3, increasing solution pH from 7 to 12 resulted in a
considerable 50% reduction in membrane fouling. This reduction of fouling led to a 53.9% decrease in the specic power consumption, while the
permeate ux increased by 97.4%. On the other hand, the mass of fouling noticeably decreased by 35.8%, the permeate ux increased by 64.5% and
the specic power consumption dramatically increased by 127.5%, when the feed ow rate was increased from 1 to 4 LPM at the solution pH of 7.
Alternatively, when the feed ow rate and the solution pH were increased simultaneously, the mass of fouling was reduced by 56.8%, with an
improvement in the specic power consumption augmented only by 54.1%, while the permeate ux increased by 136.84%.
Keywords: membrane fouling, ultrafiltration, attachments, power consumption, pH, repulsion

INTRODUCTION

n the manufacturing of paint products, cleaning of reactors and

mixing basins generate a large quantity of wastewater. Paint
efuents typically have high levels of biological oxygen
demand (BOD of >580 mg/L), chemical oxygen demand (COD
>5500 mg/L), suspended solids and turbidity.[1] Therefore, the
wastewater needs be treated before being discharged. One of the
benets of treating the waterbased paint efuent with UF over
other available processes is the potential to reclaim the solid from
the paint efuent for recycling purposes, and the possible reuse of
the permeate as processed for cleaning water.[13] However,
membrane fouling is one of the major factors that hinders a more
widespread application of UF.
Membrane fouling leads to a signicant increase in hydraulic
resistance, manifested as the decline in permeate ux or increase in
the transmembrane pressure (TMP), when the process is operated
under constantTMP or constantux conditions, respectively.[4,5]
In a system where the permeate ux is maintained by increasing
TMP, the energy required to achieve ltration escalates and causes
higher power consumption. Notably, in cases with a longer period
of ltration, membrane fouling is not completely reversible by
backwashing.[6] In fact, fouling leads to a higher operational cost,
higher energy demand, reduced membrane lifetime and increased
cleaning/washing frequency. In a seawater desalination plant the
cost of the membrane unit is about 2025% of the total capital
cost.[7] Various researches on the subject indicate that the
implementation of different fouling remediation techniques results
in a signicant reduction in the power consumption in ltration
process.[79] For a plant capable of producing 100 m3/day water,
the energy consumption is more than 10 kWh/m3; however, this
can be reduced to 3.5 kWh/m3 if fouling is being adequately
controlled.[10] The energy consumption of a crossow system is
about 5 kWh/m3 permeate. For deadend ltration the energy
consumption is only 0.10.2 kWh/m3 permeates.[11] One of the
research studies shows that the total energy consumed for 40 min
ltration was from 50.08 to 62.54 kJ/L at 3 bar while the energy

VOLUME 9999, 2014

consumed in membrane cleaning ranged from 87.08 to 107 kJ for

50100 min at different operational pressures.[12] Therefore,
innovative approaches to fouling remediation are necessary for
the reduction of power consumption and for future membrane
development and its application.
The question underlying membrane fouling and its practical
implementation is the origin of the attachment of foulants on the
membrane surface. Two major forces contributing to this foulant
attachment are dispersion interaction force and polar interaction
force.[13] Foulants that remain attached to the membrane surfaces
are most likely caused by the balance of the van der Waals
attraction force and the electrostatic repulsion force between
particles and the membrane surface due to the surface charge. Van
Oss postulated the concepts of apolar and polar interactions in
order to classify and predict interactions in an aqueous medium,
based on the Derjaguin, Landau, Verwey and Overbeek (DLVO)
theory.[13] The apolar interaction is mainly a dispersion interaction. On the other hand, the polar interaction is the interaction
between Lewis acidbase pairs in the system, including the two
interacting entities and the surrounding water molecules. These
interactions are useful in explaining the benet of hydrophilising
the membrane surface to reduce the irreversible attachment of
particles and other fouling materials on the membrane surface. The
strength of chemical attachment depends on the dispersion
interaction (apolar), as well as, predominantly on the polar
interaction. The latter can be either attractive or repulsive,
depending on the hydrophilicity of the two interacting
surfaces.[13,14]

*Author to whom correspondence may be addressed.

Email address: amira.abdelrasoul@ryerson.ca
Can. J. Chem. Eng. 9999:113, 2014
2014 Canadian Society for Chemical Engineering
DOI 10.1002/cjce.21974
Published online in Wiley Online Library
(wileyonlinelibrary.com).

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

The ionic strength of the feed solution signicantly alters the

surface properties and interactions of the colloidal foulants, and
therefore, affects the fouling potential.[15,16] For hydrophilic
membranes, membrane fouling decreased when the ionic strength
of the feed solution was amplied by increasing the solution pH.
This was attributed to the augmentation of the electrostatic
repulsion forces between the foulants and the membrane surface,
due to the increased negative charge on the membrane surface.[17
24]
The inuence of the ionic strength on fouling can be directly
observed by performing ltration experiments. The experimental
observation, however, may not be sufcient for a thorough
understanding of the fouling potential of latex at different
solution pH. The relationship between the solution ionic strength
and the particletoparticle and particletomembrane attachment
is necessary in elucidating the underlying factors affecting
membrane fouling. In our previous work a mathematical model
for ultraltration was developed.[25] The model can predict the
mass of the fouling retained by the membrane and the increase in
the transmembrane pressure. However, the developed model
expressions are dependent on the particletoparticle and particle
tomembrane attachment probabilities. Therefore, the inuence of
the ionic strength, via varying the solution pH, on the attachment
probability, and hence, on membrane fouling was investigated in
the present study. The inuence of the operational conditions on
the total mass of fouling, and the specic power consumption was
also investigated.
ATTACHMENT MATHEMATICAL MODEL
In our previous work, a mathematical model was developed to
predict the amount of fouling based on analyses of particle
attachments. The mathematical model was combined with the
Hermia model to further develop models for transmembrane
pressure prediction. Fouling was considered primarily due to the
attachments among foulant entities (coagulation attachment) and
between foulants and membrane surfaces (depositional attachment). The coagulation attachment is the particletoparticle
attachment (app) and the depositional attachment is the particle
tomembrane attachment (apm). Notably, the two attachment
probabilities are dependent on the properties of the foulants and
membranes, as well as the solution chemistry and operating
conditions. Based on the mathematical model expressions,[25] the
mass of particles contributing to pore blocking can be calculated
from Equation (1). The mass of particles accountable to the cake
layer can be calculated from Equation (2). These equations are
functions of the membranes physical properties, particle projected
area, feed concentration, volumetric permeate ux and attachment
fouling probabilities. The summation of mp and mc represents the
total mass of fouling retained by the membrane (mt). Equations (1)
and (2) are solved for app and apm using mp and mc values
measured experimentally, as described in Experimental Method
section. Furthermore, the Response Surface Methodology (RSM)
was used to design experiments that provided sufcient data and
allowed to establish the necessary correlations for the prediction of
the attachment fouling probabilities at a given operational
condition. These correlations could be then used in combination
with the mathematical model in order to make a complete set of
predictive models, as illustrated in our previous work.
Equation (3) represents the increase in the transmembrane
pressure due to membrane fouling.[25] This equation is a function
of the membranes physical properties, particle projected area, feed
concentration, volumetric permeate ux, cake resistance and
attachment fouling probabilities. The equation indicates that the

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

coagulation and depositional attachments have an effect on the

increase transmembrane pressure. Consequently, that increase in
transmembrane pressure will result in a signicant increase in the
power consumption through the ltration process
apm es
apm es
1  exp4app s L C in V s 
1  exps s C in V s 
4app s L
s


 s
 
Nm
s xs apm B

pDm Lm Bapm  ln 1 exp pDm Lm Bapm 

C in V s
Nm
s xs apm B

mp

1
mc app C in V s



apm 1  es  app 1  exp4app s L C in V s  1  exp4app s s C in V s 

app
4
sL
ss

2
1
1  apm es =4app 1  exp4app s L C in V s 
1

1  apm es 1  exps s C in V s 
1

2
m Lm Bapm 
g
f1  N m =s xs rapm BLm fln1exppDm LmpD
Bapm s xs apm B=N m C in V s g
^c 
apm es
R

C in V s  app
1  exp4app s L C in V s 
Rm
4 sL

P0 1


apm es
Nm
1  exps s C in V s  
pDm Lm Bapm
ss
s xs apm B




s xs apm B
ln 1 exp pDm Lm Bapm 
C in V s
Nm

where mp (kg/m2) is the total mass of particles attached to

membrane pores per unit membrane surface area, mc (kg/m2) is
the total mass of particles in the cake layer per unit membrane
surface area, P0 is the increase in transmembrane pressure through
the ltration time, es is the cake porosity, sL (m2/kg) is the
projected area of a unit mass of the large particles (particle
diameter  pore diameter) on membrane surface, ss (m2/kg) is the
projected area of a unit mass of the small particles on the
membrane surface (pore diameter/6 < particle size < pore diameter/2), sxs (m2/kg) is the projected area of a unit mass of the very
small particles on the membrane surface (particle size < pore
diameter/6) (At each particles size range, the average of particles
size was used for the calculations of projected area of the particles
on the membrane surface), Nm is the number density of membrane
pores per unit membrane surface area, Lm (m) is the length of a
membrane pore, Dm (m) is the membrane pore diameter, B (m2)
is the mass transfer coefcient, Cin (kg/m3) is the mass
concentration of particles at the inlet of a membrane pores, Rm
^ c (m/kg) is the cake resistance
(m1) is the membrane resistance, R
and Vs (m3/m2) is the cumulative volume of the permeate
normalised to membrane surface area.

SPECIFIC POWER CONSUMPTION

During the ltration, the resistance to the permeate ow increases
due to pore blockage and cake layer resulting in the membrane
fouling; hence, the permeate ux declines with ltration time.
Higher permeate ux can be achieved by increasing the
transmembrane pressure, which induces a higher energy

VOLUME 9999, 2014

consumption. The specic power consumption per unit volume of

ltrate is dened as:
Specific power consumption kWh=m3 1:916  106

TMP AVG Q
s
V
4

where TMP AVG is the timeaveraged transmembrane pressure

throughout the ltration duration. TMP AVG (psig min) can be
calculated based on the area under the curve, as shown in Figure 1.
 s (m3) is the cumulative
Q (LPM) is the feed ow rate and V
permeate volume.
MATERIALS AND METHODS
Experimental Method
A schematic diagram of the experimental setup is shown in
Figure 2.[25] Further details about the experimental setup, membrane
ltration unit, latex paint used and the particulars of the procedure,
were published in the previous study.[25] Samples of the feed and the
permeate streams were collected during each experimental run. A
measured mass of each sample was then transferred to a metal dish
that was placed in an oven at 1058C for the duration of a week. The
mass of each dish was periodically measured, approximately every
3 days, until there was no discernible change. From the measured
masses of the solution and the solid in each metal dish, the solid
content of the paint solution was then determined. The particle size
distribution of the latex paint solution was determined using a
particle size analyser (Model 2000; Malvern Mastersizer, Worcestershire, UK) which had a range of 0.022000 mm. The particle size
distribution of the latex solution had a range of 0.03800 mm. The
concentration of samples did not affect the particle size distribution
obtained during this experiment.
The total mass of fouling (mt) is the difference in weight of the
membrane before and after ltration. For measuring the mass of
particles contributing to the cake layer (mc), the cake was removed
(scratched off) under the microscope, so as to keep the membranes
pores blocked with the particles contributing to the pore blocking.
The scratched off solid portion was then weighted to determine the
amount of the cake layer. The blocked pores of the membrane were
checked using SEM. The mass of particles contributing to pore
blocking (mp), is the difference between the weight of the
membrane before ltration (clean membrane) and the weight
after the scratching off the cake layer. The mass of particles (mt, mc

and mp) was measured using an electronic balance (Mettler Toledo

Model AB 54S Fact, Switzerland, 0.1 mg).
The solution pH was adjusted using a pH transmitter
(Mettlertoledo pH Transmitters 2100e, MettlerToledo GmbH,
Urdorf, Germany, <0.02 pH). The solution acidity was increased
using 0.1 N H2SO4 or 0.1 N HCl in order to investigate the effects of
the divalent and monovalent acids on the membrane fouling
attachments. Similarly, 0.1 N NaOH was used to increase the
alkalinity of the latex solution. Experiments were carried out in
the pH value range between 3 and 12. The conductivity of the latex
solution (m S/cm) was measured at each pH value using a
conductivity meter (Model 4063 Traceable1 Portable Conductivity
Meter, Control Company, Friendswood, TX 0.3%). The zeta
potential (mV) of the latex particles and membrane surface was
measured at each pH value using a zeta potential analyser (Model
90 Plus zeta potential analyser, Brookhaven Instruments, Holtsville, NY, 0.01 mV). In order to obtain images of the membrane
surface before and after ltration, a scanning electron microscope
(Model JEOL, JSM6380 LV, Oxford Instrument, Abingdon, UK)
was used. A clean membrane sample was coated with a thin layer
of gold particles to improve the resolution of the images.
Polycarbonate membranes with nominal pore size of 0.05 microns
(GE Water & Process Technologies) were used for the experimental
work. Table 1 shows the membrane characteristics.
Experimental Design
The investigated operational process parameters were the transmembrane pressure, feed ow rate and feed concentration. The
temperature of the solution was kept constant at room temperature
(22248C). The operation time for each experiment was maintained constant (25 min) to investigate the inuence of the
operating conditions on fouling attachments probabilities, mass
of particles retained by membrane, volumetric permeate ux and
the specic power consumption. The ranges of the operating
conditions are shown in Table 2.
In this study, the Central Composite FaceCentered CCF
Response Surface Method (RSM) was selected as the experimental
design method.[26] Total mass of fouling and the specic power
consumption are the main responses. The process parameters
include transmembrane pressure, feed ow rate and feed
concentration. Table 3 presents coded and actual levels of the
process parameters. The coded values are designated as 1 (low), 0
(medium) and 1 (high). Twenty experiments were performed and
their results were then analysed using multiregression. For the
minimisation of error, six replicated experiments at zero code for
each process parameter were performed in randomised order.
Statistical software (StatEase, Version 8.0 StatEase Inc.) was used
to determine the besttting model by regression and stepwise
elimination. Coefcients of a full model were evaluated by
regression analysis and tested for their signicance. The signicance of the coefcients was assessed by Ftest, and the
insignicant coefcients were excluded. Pvalue analysis was
used to set the level of condence for the Ftest. The model tting
was evaluated by the coefcients of determination (R2) and the
analysis of variances (ANOVA).
RESULTS AND DISCUSSION
Effect of Operating Conditions on Fouling Attachments and
Power Consumption

Figure 1. Area under the curve represents TMP AVG (psig min).

VOLUME 9999, 2014

Figure 3 presents the effect of the transmembrane pressure on

fouling attachments, the volumetric permeate ux, the total mass

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

Figure 2. Schematic diagram of experimental setup.[25]

Table 1. Properties of GE membranes

Membrane material Pore sizea (mm) Pore densityb (pore/cm2) Nominal weight (mg/cm2) Nominal thicknessc (mm) Typical flow rated (mL/min/cm2)
Polycarbonate

0.05

6  108

0.7

0.4

Tolerance 0%, 20%.

b
Tolerance 15%.
c
Tolerance 10%.
d
Initial ow rates measured using preltered water at 10 psi.
a

of fouling and the specic power consumption over the ltration

process. Each point on the curves in the gures represents the result
of an individual experimental run. As shown in Figure 3a, at a feed
ow rate of 4.0 L/min and a feed solid concentration of 1.3 kg/m3,
when the pressure was raised from 15 to 35 psig, the particleto
particle (app) and the particletomembrane (apm) attachment
probabilities increased from 0.4 to 0.75 and from 0.5 to 0.8,
respectively. However, further increase in pressure did not affect
app and apm signicantly. When the transmembrane pressure was
raised from 15 to 35 psig, the volumetric permeate ux increased
from 0.0125 to 0.0178 m3/m2. Higher permeate ow through the
membrane brought more particles to the membranes surface. As a
consequence, the chances for the particletoparticle and particle
tomembrane collision and attachment increased, resulting in

higher attachment probabilities. Notably, increasing the pressure

from 35 to 45 psig resulted in a insignicant change in the
volumetric permeate ux (Vs) from 0.0178 to 0.0179 m3/m2. This is
due to the fact that the transmembrane pressure higher than the
critical pressure (35 psig) has an insignicant effect on the
permeate ux or the total mass of fouling retained by the
membrane. Thus, the fouling attachments probabilities values
levelled off to a relatively constant value. Moreover, Figure 3a
shows that the depositional attachment probability apm was higher
than the coagulation attachment probability app. This indicates that
the transmembrane pressure might have a more signicant impact
on the pore blocking and the membranetoparticle attachment
than the cake layer buildup. Higher fouling attachment probabilities resulted in an increase in the total mass of fouling from 0.0104

Table 2. Process parameters levels

Range
Factor
A
B
C

Variable

Low (1)

High (1)

Unit

Transmembrane pressure
Feed flow rate
Feed concentration

15
1
0.78

35
7
1.82

psig
LPM
kg/m3

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

VOLUME 9999, 2014

Table 3. Experimental parameters and results based on the experimental design for the total mass of fouling and the specic power consumption
Independent variables

Exp no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
a

Initial TMP
(psig)
15
35
15
35
15
35
15
35
15
35
25
25
25
25
25
25
25
25
25
25

(1)
(1)
(1)
(1)
(1)
(1)
(1)
(1)
(1)
(1)
(0)
(0)
(0)
(0)
(0)
(0)
(0)
(0)
(0)
(0)

TMPAVGa
(psig)

Final TMP
(psig)

15
35
15
35
16.47
35.27
16.27
35
15
35
25
25
25
26.8
25.5
25
25
25
25
25

17
40
16
39
20
42
21
43
18
41
32
27
27
33
29
29
29
29
29
29

Feed flow rate Q

(LPM)
1
1
7
7
1
1
7
7
4
4
1
7
4
4
4
4
4
4
4
4

(1)
(1)
(1)
(1)
(1)
(1)
(1)
(1)
(0)
(0)
(1)
(1)
(0)
(0)
(0)
(0)
(0)
(0)
(0)
(0)

Feed concentration Cf
(kg/m3)
0.78
0.78
0.78
0.78
1.82
1.82
1.82
1.82
1.3
1.3
1.3
1.3
0.78
1.82
1.3
1.3
1.3
1.3
1.3
1.3

(1)
(1)
(1)
(1)
(1)
(1)
(1)
(1)
(0)
(0)
(0)
(0)
(1)
(1)
(0)
(0)
(0)
(0)
(0)
(0)

mt
(kg/m2)

Power consumed
(kWh/m3)

0.0088
0.013
0.0048
0.012
0.018
0.019
0.0155
0.023
0.01
0.015
0.0195
0.0082
0.0076
0.024
0.012
0.0117
0.0118
0.012
0.0121
0.0119

2.81
5.74
14.74
28.46
4.82
8.66
25.98
37.97
10.17
16.61
5.34
19.64
12.01
17.23
13.98
13.77
13.61
13.84
13.66
13.51

TMPAVG (psig) TMP AVG (psig min)/ltration time (min), TMP AVG (psig min) calculated as mentioned in Specic Power Consumption section.

permeate ux (Vs) increased from 0.0076 to 0.014 m3/m2. As a

result, more particles were brought into contact with the
membrane, leading to an increase in collisions and attachments
between particles and the membrane surface. Consequently, apm
decreased at a lower rate as compared with that of app. As shown in
Figure 4a, app and apm approach the highest values at ow rate of
1.0 LPM. It was also observed that as the fouling attachment
probabilities decreased, the mass of particles retained by the
membrane dropped from 0.0162 kg/m2 at a feed low rate of 1 LPM
to 0.00838 kg/m2 at 6 LPM, as illustrated in Figure 4b. Due to the
reduction of the total mass of fouling by increasing the feed ow
rate from 1 to 6 LPM, the transmembrane pressure over the
duration of the ltration process only increased by 1 psig,
compared to an increase of 5 psig at 1 LPM. Although the total
mass of fouling was decreased, with the increase in the ow rate
from 1 to 6 LPM, the power consumption per unit volume increased
from 4.47 to 13.1 kWh/m3 as illustrated in Figure 4b. This aligned

to 0.0145 kg/m2 as shown in Figure 3b. The specic power

consumption also increased from 10.17 to 21.02 kWh/m3.
Figure 4a depicts the effect of the feed ow rate on the fouling
attachment probability at the initial transmembrane pressure of
15 psig and the feed solid concentration of 1.3 kg/m3. Under these
conditions, the particletoparticle attachment probability (app)
was signicantly reduced from 0.86 at 1 LPM (crossow velocity
of 10.4 cm/s) to 0.19 at 6 LPM (crossow velocity of 62.5 cm/s)
while the particletomembrane attachment probability (apm)
decreased from 0.82 to 0.49. The probable cause for this substantial
reduction in app with the feed ow rate is that at a higher ow rate,
a higher shear rate is applied on the particles at the cake layer.
Consequently, a higher number of particles were washed away
from the cake layer. In particular, as the feed ow rate was
increased to 3.0 LPM, there was a signicant decrease in app, which
arguably marks the crucial ow rate for membrane fouling in this
system. On the other hand, at a higher ow rate, the cumulative

Figure 3. Effect of initial transmembrane pressure at [Q 4 LPM], [Cf 1.3 kg/m3], [pH 7] on (a) fouling attachment probabilities (app, apm) and the
volumetric permeate ux (Vs). (b) Total mass of fouling and specic power consumption.

VOLUME 9999, 2014

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

Figure 4. Effect of feed ow rate at [TMP 15 psig], [Cf 1.3 kg/m3] and [pH 7] on (a) fouling attachment probabilities (app, apm) and the volumetric
permeate ux (Vs). (b) Total mass of fouling and power consumption.

with the specic power consumption as directly proportional to the

feed ow rate in Equation (4). Specically, this indicates that the
increasing of the ow rate six times results in a three times
increasing of the power consumption. This, in turn, implies that the
remediation of the fouling is greatly required to reduce the total
mass of fouling in lower ow rates, increase the volumetric
permeate ux, and, at the same time, reduce the power
consumption per unit volume of ltrate.
Figure 5a shows the effects of the feed solid concentration on the
fouling attachment probability at a feed ow rate of 1 L/min and
15 psig. Increasing the feed concentration from 0.78 to 1.82 kg/m3
resulted in a signicant increase in the coagulation attachment
probability, app, and a substantial decrease in the depositional
attachment probability, apm. The particletoparticle attachment, app,
increased from 0.3 to 0.9, while the particletomembrane attachment, apm, decreased from 0.95 to 0.6. At a higher feed concentration,
the chance of particletoparticle collision and attachment would be
higher. As a consequence, the coagulation attachment probability
and cake layer buildup were enhanced. On the other hand,
increasing of the feed concentration lead to a decrease in the
volumetric permeate ux (Vs) from 0.011 to 0.007 m3/m2. A reduced
permeate ow resulted in a lower chance of particletomembrane
collision and attachment. The depositional attachment probability,
apm, thus decreased from 0.95 to 0.6. Figure 5b indicates that the total
mass of fouling increased from 0.0088 to 0.0188 kg/m2 when the
concentration was increased from 0.78 to 1.82 kg/m3, as expected.
Consequently, the increase in transmembrane pressure over the
duration of ltration was elevated from 2 to 5 psig to overcome the
higher resistance to the permeate ow; hence, the specic power
consumption was also almost doubled from 2.8 to 4.8 kWh/m3.

Statistical Analysis
Central Composite FaceCentered CCF Response Surface Method
(RSM) was selected as the experimental design method as
illustrated in details in Experimental Method section. The total
mass of fouling was measured for each experimental run, and the
specic power consumption was calculated, as shown in Table 3
using Equation (4). The total mass of fouling and the power
consumed per unit volume of permeate varied in the ranges of
0.00480.024 kg/m2 and 2.8137.97 kWh/m3, respectively.
The results were analysed using multiregression so as to identify
the effects of the process parameters on the total mass of fouling
and the specic power consumption in ultraltration of a latex
solution. Moreover, the process parameters were examined in
order to identify the optimum conditions that would allow for the
minimum mass of fouling and power consumption.
Based on the ANOVA analysis, linear models best tted the
experimental data for the total mass of fouling and the specic
power consumption. The ANOVA result is presented in Table 4.
The Pvalue is the probability value used to set the level of
condence for the Ftest and to ascertain the signicance of each of
the coefcients. The Pvalue should be less than or equal to 0.05 to
be signicant.[27,28] According to regression coefcients and
probability values (Pvalue), as shown in Table 4, the linear terms
were all functionally signicant for the total mass of fouling and the
specic power consumption. The models for the total mass of
fouling and the specic power consumption are given in
Equations (5) and (6), respectively. In addition, the goodness of
t was determined by evaluating the coefcients of determination
(R2). R2 was 0.97 for the linear models

Figure 5. Effect of feed concentration at [Q 1 LPM], [TMP 15 psig] and [pH 7] on (a) fouling attachment probabilities (app, apm) and the volumetric
permeate ux (Vs). (b) Total mass of fouling and power consumption.

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

VOLUME 9999, 2014

Table 4. Regression coefcients and probability values of statistical analysis

Coded factor
A
B
C

Actual factor

Coefficient for mt
(kg/m2)

Pvalue

Coefficient for specific

power consumption (kWh/m3)

Pvalue

TMPAVG (psig)
Feed flow rate (LPM)
Feed concentration (kg/m3)

0.000299
0.000326667
0.00928846

0.0044
0.0005
<0.0001

0.39780
3.34267
5.77692

0.0005
<0.0001
0.0045

Values of Pvalue <0.05 indicate model terms are signicant.

In case of total mass of fouling and specic power consumption models: A, B and C are signicant model terms.

Total mass of fouling mt kg=m2

0:00499 0:000299A  0:0003267B 0:00929C

Specific Power Consumption kWh=m3

16:2412 0:3978A 3:3427B 5:777C

Effect of pH on ionic strength of latex solution and membrane

surface hydrophilicity

where the total mass of fouling (mt) and the specic power
consumption are the function of average transmembrane pressure
TMPAVG (A) (psig), feed ow rate (B) (LPM) and feed concentration (C) (kg/m3). The linear models were compatible with the
results in Effect of Operating Conditions on Fouling Attachments
and Power Consumption section. As illustrated in Equation (5), in
order to minimise the total mass of fouling, the transmembrane
pressure and feed concentration should be decreased while the feed
ow rate should be increased. On the other hand, to minimise the
specic power consumption, the transmembrane pressure, feed
ow rate and feed concentration should be decreased. Furthermore
the average transmembrane pressure TMPAVG (psig) can be
estimated so as to predict the specic power consumption from
Equation (6) using the initial transmembrane pressure, variation of
transmembrane pressure (P0 ) with time estimated from Equation (3) and the ltration time.

The ionic strength is directly proportional to the solution

conductivity. Therefore, the effect of the solution pH on the ionic
strength of the solution is evaluated based on the change in the
solution conductivity with pH. Figure 6a illustrates the effect of pH
on the conductivity of the latex solution. At a solid concentration of
1.3 kg/m3, the pH of the latex solution was 7. The solution
conductivity increased from 0.094 to 38.8 m S/cm when the
solution pH was increased from 7 to 14. On the other hand,
decreasing pH from 7 to 2 using HCl and H2SO4 only resulted in a
slight increase in the conductivity to 2.57 and 5.5 m S/cm,
respectively. Figure 6b depicts the zeta potential values of the
latex particles and membrane surface. Zeta potential of latex
particles at pH 7 is approximately 26.61 mV. For the pH values
>7, the zeta potential becomes more negative as the ionic strength
increases. The zeta potential of latex particles achieved 42.66 mV
at pH 12.
The zeta potential of the membrane surface is negative at all pH
values investigated. As the pH is increased from 3 to 12, the zeta
potential of the membrane surface becomes increasingly negative
from 18.11 to 46.77 mV, respectively. This indicates that the
hydrophilicity of the membrane surface increased with the rise
of pH value.

Fouling Remediation by Changing the Solution pH

Effect of pH change on membrane fouling and power consumption

Following previous results, the fouling remediation can be

considered critical in reducing membrane fouling, increasing the
permeate ux and reducing the power consumption per unit
volume of ltrate at given operating conditions. The effects of the
solution pH over a range from 3 to 12 on particleparticle
attachment and particlemembrane were investigated since
those attachment potentials strongly affect the volumetric
permeates ux, the total mass of fouling and the specic power
consumption.

As previously shown in Effect of Operating Conditions on Fouling

Attachments and Power Consumption section and Statistical
Analysis section, at the initial transmembrane pressure of
15 psig, the feed ow rate of 4 LPM and the feed solid concentration
of 1.3 kg/m3, the power consumption per unit volume of ltrate
was 10.17 kWh/m3; and the volumetric permeate ux and the total
mass of fouling were 0.0125 and 0.0104, respectively. Under the
same operating conditions, increasing the solution pH from 7 to 12
resulted in a signicant reduction in the particletoparticle

Figure 6. The effect of pH on (a) conductivity of latex solution and (b) zeta potential of latex particles and membrane surface.

VOLUME 9999, 2014

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

attachment from 0.4 to 0.19, as shown in Figure 7a. This indicates

that the ionic strength of the latex solution increased with pH,
which in turn caused an increased negative charge on the latex
particles due to the OH group added to the solution. Raising
the pH from 7 to 12 resulted in an increase of the zeta potential of
the latex particles from 26.61 to 42.66 mV, respectively. As a
consequence, the electrostatic repulsive forces between the latex
particles were increased, while the particletoparticle attachment
decreased signicantly. Also, the depositional attachment decreased from 0.55 to 0.26. This could be attributed to the
enhancement of the hydrophilicy of the membrane and the
increase in the negative charge on the membrane surface with
increases in the solution pH. As the pH is increased from 7 to 12, the

zeta potential of the membrane surface became increasingly

negative from 36.67 to 46.77 mV, respectively. Therefore, the
electrostatic attraction force between latex particles and the
membrane surface decreased. Consequently, the particleto
membrane attachment was signicantly reduced. The noticeable
reduction of the fouling attachments caused a considerable
decrease in the total mass of fouling from 0.0104 to 0.007 kg/m2,
as can be seen in Figure 7b. This indicates that an increase in the
ionic strength via an increase in pH resulted in some enhancement
of the antifouling property of the feed and the membrane. The
decrease in the total mass of fouling caused a reduction in the
resistance to the ow. Therefore, the increase in the transmembrane pressure over the duration of the ltration process was

Figure 7. Effect of pH value at [TMP 15 psig], [Cf 1.3 kg/m3] and [Q 4 LPM] (using 1 N NaOH and 0.1 N H2SO4) on (a) fouling attachment probabilities
(app, apm) and the volumetric permeate ux (Vs). (b) Total mass of fouling and power consumption (kWh/m3).

Figure 8. SEM images of membrane after ultraltration [Q 4 LPM], [Cf 1.3 kg/m3], [TMP 15 psig] (a) cake height 5.05 mm at pH 7, (b) cake
height 3.6 mm at pH 10 and (c) cake height 1.85 mm at pH 12.

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

VOLUME 9999, 2014

reduced, and hence, the specic power consumption decreased

from 10.17 to 6.9 kWh/m3. In addition, decreasing of both
attachments resulted in increasing permeate ux from 0.0125 to
0.018 m3/m2, as shown in Figure 7a.
On the other hand, in acidic conditions the hydrophilic
membrane became less negatively charged (more hydrophobic).
This can be attributed to the more solubility of H functional
group. As the pH decreased from 7 to 3, the zeta potential of the
membrane surface became signicantly less negative, decreasing
from 36.67 to 18.11 mV, respectively. This in turn results in a

stronger electrostatic attractive force between the latex particles

and the membrane surface. As a consequence, more particleto
membrane attachment would occur. As illustrated in Figure 7a,
when pH was decreased from 7 to 3, a considerable increase in the
depositional attachment from 0.55 to 0.82 was observed.
Moreover, the lowering of the pH reduced the negative charge
on the particles, thus causing a stronger electrostatic attraction
force between particles and encouraging particle aggregation. As
the pH decreased from 7 to 3, the zeta potential of the latex particles
became signicantly less negative, decreasing from 26.61 to

Figure 9. Effect of pH value at [TMP 15 psig], [Cf 1.3 kg/m3] and [Q 1 LPM] (using 1 N NaOH and 0.1 N H2SO4) on (a) fouling attachment probabilities
(app, apm) and the volumetric permeate ux (Vs). (b) Total mass of fouling and power consumption (kWh/m3).

Figure 10. SEM images of membrane after ultraltration. (a) Cake height 13.8 mm at [TMP 15 psig], [Cf 1.3 kg/m3], [Q 1 LPM] and pH 7. (b) Cake
height 6.1 mm at [TMP 15 psig], [Cf 1.3 kg/m3], [Q 1 LPM] and pH 10. (c) Cake height 2.3 mm at [TMP 15 psig], [Cf 1.3 kg/m3], [Q 1 LPM] and pH
12. (d) Cake height 2.4 mm at [TMP 15 psig], [Cf 1.3 kg/m3], [Q 6 LPM] and pH 7.

VOLUME 9999, 2014

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

permeate ux of 0.015 m3/m2 was obtained, which is higher than

the permeate ux at a higher feed ow rate of 6 LPM with the
solution pH of 7 and the same other operating conditions.
Furthermore, the cake height at Q 1 LPM, TMP 15 psig,
Cf 1.3 kg/m3 and pH 7 was 13.8 mm as shown in Figure 10a.
Increasing pH value to 10 resulted in a signicant reduction in the
cake thickness to 6.1 mm. At pH 12 the cake thickness was further
reduced to 2.3 mm, which is very close to the cake thickness of
2.4 mm at Q 6 LPM, TMP 15 psig, Cf 1.3 kg/m3 and pH 7.
However, at the lower feed ow rate of 1 LPM, TMP 15 psig,
Cf 1.3 kg/m3 and pH 12, the specic power consumption was
only 2.06 kWh/m3 (Figure 9b), compared with 13.1 kWh/m3 at
Q 6 LPM, TMP 15 psig, Cf 1.3 kg/m3 and pH 7 (Figure 4b).
Table 5 summarises the difference between the effects achieved
by either increasing the feed ow rate or by raising the pH value
with respect to the fouling attachment, the total mass of fouling, the
volumetric permeate ux, and the specic power consumption at
transmembrane pressure of 15 psig and 1.3 kg/m3. A sole increase
of the feed ow rate reduced membrane fouling, but was
accompanied by high power consumption. On the other hand,
an increased feed ow rate accompanied by an augmented pH
value had a noticeable improvement in the specic power
consumption. On the overall, a reduced ow rate in combination
with an increased pH had the best effects on both the membrane
fouling and the specic power consumption.
The effects of ionic strength on the membrane fouling and the
power consumption were also studied at a higher TMP of 25 psig,
Q 4 LPM and Cf 1.3 kg/m3. The solution pH was varied within a
range from 3 to 12. Figure 11 shows similar trends for the fouling
attachments, the volumetric permeate ux, the total mass of
fouling, and the specic power consumption as those at TMP 15

11.2 mV, respectively. A stronger electrostatic attraction and

particle aggregation resulted in an increased coagulation attachment from 0.4 to 0.59. Moreover, SEM images obtained after the
ultraltration provided further evidence in support of the results
acquired previously. Particularly, at pH 7 the cake thickness was
5.05 mm, as shown in Figure 8a. The cake height decreased to
3.6 mm at pH 10, as can be seen in Figure 8b. The cake thickness
further diminished to 1.85 mm at pH 12, as illustrated in Figure 8c.
This was due to the signicant reduction in the particletoparticle
attachment, which was the predominant factor responsible for the
cake formation.
As presented in Effect of Operating Conditions on Fouling
Attachments and Power Consumption section and Statistical
Analysis section, the minimum specic power consumption
occurred at the lowest ow rate of 1 LPM. On the other hand, at
the lowest ow rate the total mass of fouling and the cake layer
thickness were highest. Therefore, the effect of pH on the
membrane fouling and the specic power consumption at a low
feed ow rate of 1 LPM was further investigated. It was observed
that at a feed ow rate of 1 LPM, the transmembrane pressure of
15 psig and the feed solid concentration of 1.3 kg/m3, when the
solution pH was increased from 7 to 12, the total mass of fouling
decreased from 0.0162 to 0.0081 kg/m2, as can be seen in
Figure 9b. This contributed to the considerable reduction in both
attachment probabilities app and apm, from 0.86 to 0.40 and from
0.82 to 0.37, respectively, as shown in Figure 9a. Therefore, the
membrane fouling was lessened, leading to a decrease in the
transmembrane pressure elevation over the duration of the
experiment from 6 to 1 psig. Consequently, the specic power
consumption decreased from 4.62 to 2.06 kWh/m3, as presented in
Figure 9b. Moreover, at the solution pH of 12, a volumetric

Table 5. Comparison of total mass of fouling, volumetric permeate ux, cake height and the power consumption at different pH. [TMP 15 psig],
[Cf 1.3 kg/m3] and 25 min
Q (LPM)

pH

app

apm

mt (kg/m2)

Vs (m3/m2)

Specific power consumption (kWh/m3)

Cake height (mm)

1
1
1
4
4
4

7
10
12
7
10
12

0.86
0.56
0.4
0.4
0.26
0.19

0.82
0.55
0.37
0.55
0.34
0.26

0.0162
0.011
0.0081
0.0104
0.008
0.007

0.0076
0.012
0.015
0.0125
0.016
0.018

4.47
2.65
2.06
10.17
7.87
6.89

13.8
6.1
2.3
5.05
3.6
1.85

Figure 11. Effect of pH value at [TMP 25 psig], [Cf 1.3 kg/m3] and [Q 4 LPM] (using 1 N NaOH& 0.1 N H2SO4) on (a) fouling attachment probabilities
(app, apm) and the volumetric permeate ux (Vs). (b) Total mass of fouling and power consumption (kWh/m3).

10

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

VOLUME 9999, 2014

Figure 12. SEM images of membrane after ultraltration [Q 4 LPM], [Cf 1.3 kg/m3], [TMP 25 psig] (a) Cake height 6.65 mm at pH 7, (b) cake height
5.30 mm at pH 10.

psig shown Figure 7. However, ultraltration at TMP of 25 psig

resulted in higher values for the particletoparticle and particleto
membrane attachments. For instance, at pH 10, the particleto
particle attachment was 0.40 and 0.26, at TMP of 25 and 15 psig,
respectively. At a higher TMP, the permeate ux was higher. More
particles were thus brought to the membrane surface; hence, more
chances for the particletoparticle and particletomembrane
collision and attachment occurred. Consequently, the particleto
particle and particletomembrane attachment increased. At TMP
25 psig and Q 4 LPM, increases in pH from 7 to 12 resulted in a
signicant reduction in the total mass of fouling from 0.012 to
0.0077 kg/m2. Consequently, the specic power consumption was
reduced from 14.0 to 10.1 kWh/m3, as shown in Figure 11b.
Figure 12a and b presents SEM images of the cake height after
ultraltration at the solution pH of 7 and 10, respectively. The cake
thickness decreased from 6.65 to 5.30 mm when the pH was raised
from 7 to 10. SEM images in Figure 12 and Figure 8 provide a
conrmation of the effect of the transmembrane pressure on the
cake height.
Effect of divalent and monovalent ions
The effect of the monovalent ion on fouling attachments, total mass
of fouling, and specic power consumption was also investigated
using sulphuric acid and hydrochloric acid. As shown in Table 6
at pH of 3 and 5, lower fouling attachment probabilities, lower total
mass of fouling, higher volumetric permeate ux and lower power
consumption were achieved with sulphuric acid. This can be
attributed to the more negative charge sulphate ion, SO42, of
sulphuric acid, which makes the membrane less hydrophobic in
comparison to the chlorine ion, Cl, of hypochloric acid. It is
relevant to note that the zeta potential of the membrane surface
at pH 3 was 18.11 and 16.31 mV with sulphuric acid and

Table 6. Comparison of the effect of the monovalent acid (HCl) and the
divalent acid (H2SO4) at pH 3 and 5 at transmembrane pressure 15 psig,
4 LPM and 1.3 kg/m3

pH 3
app
apm
mt (kg/m2)
Vs (m3/m2)
Power consumption (kWh/m3)
pH 5
app
apm
mt (kg/m2)
Vs (m3/m2)
Power consumption (kWh/m3)

Using
0.1 N H2SO4

Using
0.1 N HCl

0.59
0.82
0.0135
0.0098
14.14

0.62
0.97
0.0145
0.0088
15.94

0.5
0.71
0.0123
0.0115
11.7

0.54
0.8
0.013
0.01
13.86

hydrochloric acid, respectively. When pH was increased to 5, the

zeta potential of the membrane surface increased to 24.78 mV
using sulphuric acid, in comparison to 20.55 mV using hydrochloric acid. This again reconrms the advantage of sulphuric acid
over hydrochloric acid in fouling mediation. Figure 13 presents the
SEM images of the cake thickness at pH 3 using sulphuric acid and
hypochloric acid. The cake thickness increased to 6.35 mm for 0.1 N
H2SO4 to 7.65 mm for 0.1 N HCl.
CONCLUSION
According to the results obtained in the present study, it was
observed that higher fouling probabilities caused an increase in the

Figure 13. SEM images of membrane after ultraltration [Q 4 LPM], [Cf 1.3 kg/m3], [TMP 15 psig], pH 3 (a) Cake height 6.35 mm using 0.1 N H2SO4.
(b) Cake height 7.65 mm using 0.1 N HCl.

VOLUME 9999, 2014

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

11

total mass of fouling retained by the membrane and the cake

thickness, which in turn resulted in an increased resistance to the
permeate ux. Higher permeate volume could be achieved by
increasing the transmembrane pressure, thus causing a greater
energy consumption per unit volume of ltrate. For example, at a
feed ow rate of 4.0 L/min and feed concentration of 1.3 kg/m3,
increase in the transmembrane pressure from 15 to 45 psig leads to
higher coagulation and depositional attachments. The total mass of
fouling also increased accordingly from 0.0104 to 0.0145 kg/m2.
Furthermore, the specic power consumption increased from
10.17 to 21.02 kWh/m3. In addition, increasing the feed concentration from 0.78 to 1.82 kg/m3 caused a substantial increase in the
total mass of fouling that is, from 0.0088 to 0.0188 kg/m2. Notably,
in this case the specic power consumption was almost doubled
from 2.8 to 4.8 kWh/m3. It was also observed that both attachment
probabilities considerably decreased when the feed ow rate was
increased from 1 to 6 LPM; resulting lower total mass of fouling and
cake height as well. On the other hand, the specic power
consumption increased signicantly from 4.47 to 13.1 kWh/m3.
It was also observed that the lowest specic power consumption
occurred at the minimum ow rate with the minimum transmembrane pressure and the lowest feed concentration. In addition,
lower membrane fouling, higher volumetric permeate ux and
lower specic power consumption were obtained with increasing
the solution pH. It was noted that at given transmembrane
pressure, feed solid concentration and pH, increasing the feed ow
rate resulted in a reduced total mass of fouling. However, the
specic power consumption increased. On the other hand, increase
in the pH from 7 to 12 lead to decrease in the total mass of fouling
and the specic power consumption. A combination of an
increased feed ow rate and a higher solution pH was thus shown
to enhance both the permeate ux and the specic power
consumption with less membrane fouling.
NOMENCLATURE
B
Cin
Cf
Dm
Lm
mp
mc
mt
Nm
P0
Pvalue
Q
R2
Rm
^c
R
TMP
TMP AVG
TMPAVG

12

mass transfer coefficient (m2)

the mean mass concentration of particles at the inlet of
a membrane pores (kg/m3)
concentration of foulants in the feedwater (kg/m3)
membrane pore diameter (m)
length of membrane pores (m)
total mass of particles attached to membrane pores per
unit area (kg/m2)
total mass of particles in the cake layer per unit area
(kg/m2)
total mass of particles retained per unit membrane
surface area (kg/m2)
number density of membrane pores per a unit
membrane surface area
the increase in transmembrane pressure due to
membrane fouling through the filtration time
probability in statistical significance testing (ANOVA
test)
feed flow rate (LPM)
the percent of the variation of the response explained
by the model
membrane resistance (m1)
Cake resistance (m/kg)
initial transmembrane pressure
timeaveraged transmembrane pressure throughout
the filtration duration (psig min)
average transmembrane pressure throughout the

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

Vs
s
V

filtration duration (psig)

cumulative volume of the permeate normalized to
membrane surface area (m3/m2)
cumulative permeate volume (m3)

Greek Symbols
s
sL
sS
sxS
es
apm
app

projected area of a unit mass of the particles on membrane

surface (m2/kg)
projected area of a unit mass of the large particles (Particle
diameter  pore diameter) on membrane surface (m2/kg)
projected area of a unit mass of the small particles on
membrane surface (m2/kg) (pore diameter/6 < particle
size < pore diameter/2)
projected area of a unit mass of the very small particles on
membrane surface (m2/kg) (particle size < pore diameter/
6)
cake porosity
the attachment probabilities between a particle and the
membrane
the attachment probabilities between two particles

ACKNOWLEDGEMENTS
The authors are grateful for the nancial support from Natural
Science and Engineering Research Council of Canada (NSERC).
The assistance and facilities provided by the Department of
Chemical Engineering, Ryerson University, are also highly
appreciated.
REFERENCES
[1] B. K. Dey, M. A. Hashim, S. Hasan, B. S. Gupta, Adv. Environ.
Res. 2004, 8, 455.
[2] B. Krbahti, N. AktaSs , A. Tanyola, J. Hazard. Mater. 2007,
148, 83.
[3] U. Kurt, Y. Avsar, M. T. Gonullu, Chemosphere 2006, 64,
1536.
[4] M. Kennedy, J. Kamanyi, S. Rodriguez, N. Lee, J. Schippers,
G. Amy, Adv. Membr. Technol. Appl. 2008, 131.
[5] M. H. Javacek, F. Bouchet, J. Membr. Sci. 1993, 82, 285.
[6] S. Kosvintsev, I. W. Cumming, R. G. Holdich, D. Lloyd, V. M.
Starov, Colloids Surf. A 2004, 230, 167.
[7] A. AlAmoudi, R. W. Lovitt, J. Membr. Sci. 2007, 303, 4.
[8] A. Hu, D. Stuckey, J. Environ. Eng. 2006, 132, 190.
[9] H. B. Winzeler, G. Belfort, J. Membr. Sci. 1993, 80, 35.
[10] S. A. Avlonitis, K. Kouroumbas, N. Vlachakis, Desalination
2003, 157, 151.
[11] A. K. Teuler, J. M. Glucina, Desalination 1999, 125, 89.
[12] A. H. Bahnasawy, M. E. Shenana, Aust. J. Agric. Eng. 2010,
15, 54.
[13] J. Israelachvili, Intermolecular & Surface Forces, 2nd
edition, Academic Press Inc., San Diego, CA 1992. p. 450
[14] J. Gregory, Particles in Water: Properties and Processes,
Taylor & Francis Group, CRC Press, UK 2005, p. 63.
[15] G. Singh, L. Song, J. Colloid Interface Sci. 2005, 284, 630.
[16] B. P. Binks, T. S. Horozov, Colloidal Particles at Liquid
Interfaces, Cambridge University Press, Cambridge 2006.
[17] R. Faibish, M. Elimelech, Y. Cohen, J. Colloid Interface Sci.
1998, 204, 77.

VOLUME 9999, 2014

[18] A. E. Childress, M. Elimelesh, J. Membr. Sci. 1996, 119, 253.

[19] S. S. Deshrnukh, A. E. Childress, Desalination 2001, 140, 87.
[20] J. Na, Z. Yonggang, The Effect of Humic Acid on
Ultraltration Membrane Fouling, in International Conference on Computer Technology and Science (ICCTS 2012),
IPCSIT Vol. 47, IACSIT Press, Singapore 2012.
[21] M. Elimelech, W. Chen, J. Waypa, Desalination 2010, 95, 269.
[22] M. Chun, H. Cho, I. K. Song, Desalination 2002, 148, 363.
[23] M. Mika, P. Arto, N. Marianne, J. Membr. Sci. 2006, 280, 311.
[24] K. L. Jones, C. R. OMelia, J. Membr. Sci. 2000, 165, 31.
[25] A. Abdelrasoul, H. Doan, A. Lohi, J. Membr. Sci. 2013, 433,
88.
[26] D. C. Montgomery, Design and Analysis of Experiment, 5th
edition, Wiley Inc., New York, USA 1997.
[27] E. Yuliwati, A. F. Ismail, W. J. Lau, B. C. Ng, A. Mataram,
M. A. Kassim, Desalination 2012, 287, 350.
[28] M. Calatayud, M. Vela, S. Blanco, J. Garca, E. Rodrguez,
Chem. Eng. J. 2010, 156, 337.

Manuscript received September 24, 2013; revised manuscript

received October 28, 2013; accepted for publication October 28,
2013.

VOLUME 9999, 2014

THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING

13