496

IWA Publishing 2012 Water Science & Technology

65.3

2012

Optimizing the coagulation process in a drinking water

treatment plant comparison between traditional
and statistical experimental design jar tests
M. Zainal-Abideen, A. Aris, F. Yusof, Z. Abdul-Majid, A. Selamat
and S. I. Omar

ABSTRACT
In this study of coagulation operation, a comparison was made between the optimum jar test values
for pH, coagulant and coagulant aid obtained from traditional methods (an adjusted one-factor-at-atime (OFAT) method) and with central composite design (the standard design of response surface
methodology (RSM)). Alum (coagulant) and polymer (coagulant aid) were used to treat a water source
with very low pH and high aluminium concentration at Sri-Gading water treatment plant (WTP)
Malaysia. The optimum conditions for these factors were chosen when the nal turbidity, pH after
coagulation and residual aluminium were within 05 NTU, 6.57.5 and 00.20 mg/l respectively.
Traditional and RSM jar tests were conducted to nd their respective optimum coagulation conditions.
It was observed that the optimum dose for alum obtained through the traditional method was 12 mg/l,
while the value for polymer was set constant at 0.020 mg/l. Through RSM optimization, the optimum
dose for alum was 7 mg/l and for polymer was 0.004 mg/l. Optimum pH for the coagulation
operation obtained through traditional methods and RSM was 7.6. The nal turbidity, pH after

M. Zainal-Abideen (corresponding author)

A. Aris
Faculty of Civil Engineering,
Universiti Teknologi Malaysia,
81310 Johor-Bahru,
Malaysia
E-mail: muzaffar@utm.my
F. Yusof
Z. Abdul-Majid
Faculty of Science,
Universiti Teknologi Malaysia,
81310 Johor-Bahru,
Malaysia
A. Selamat
S. I. Omar
SAJ Holdings Sdn. Bhd.,
P.O. Box 262,
80350 Johor-Bahru,
Malaysia

coagulation and residual aluminium recorded were all within acceptable limits. The RSM method was
demonstrated to be an appropriate approach for the optimization and was validated by a further test.
Key words

| coagulation, jar test, one-factor-at-a-time, response surface methodology,

water treatment

INTRODUCTION
Coagulation is one of the most important processes in water
treatment. It is an effective method for the removal of colloidal particles in surface water. Many researchers have
applied the coagulation process to treat highly turbid
water (Lin et al. ) and natural organic matter (NOM)
(Zhan et al. ) present in surface water. The process is
also capable of removing arsenic (Hering et al. ) and
residual aluminium (Brub & Dorea ) from drinking
water. As a coagulation process has the ability to eliminate
many pollutants from surface and drinking water, the success of the process has a direct impact on the reliability of
treatment plant operations and nal water quality.
The effectiveness of the coagulation process is
highly dependent on many factors, including dosage of coagulant and coagulant aids and also pH of the operation
(Amirtharajah & OMelia ). By far, the most common
doi: 10.2166/wst.2012.561

coagulants used are aluminium sulphate (Al2(SO4)3), ferric

sulphate (Fe2(SO4)3) and ferric chloride (FeCl3) (Jiang &
Graham ). Studies on the performance of polymerized
inorganics such as polyaluminium chloride (PACl) (Lin
et al. ) and polyferric chloride (PFC) (Zhan et al. )
as coagulants are actively being carried out. The impact of
pH towards coagulation has also been thoroughly investigated (Gregory & Carlson ; Brub & Dorea ;
Zhan et al. ).
It is well established that the choice of coagulants used,
the dosing and the operational pH applied in a coagulation
process make a signicant contribution to the operational
cost of the treatment plant. Therefore, it is important to
use the optimum conditions when carrying out coagulation
process so that wastage or unnecessary dosage of the associated chemicals may be prevented.

497

M. Zainal-Abideen et al.

Optimizing coagulation process through traditional & RSM jar test

Several methods have been adopted to determine the

optimum coagulation conditions such as zeta-potential
measurement (Morfesis et al. ) and jar tests. Of these
methods, the jar test has been commonly used to determine
the best pH and dosage of the chemicals owing to its simplicity. In a typical jar test practice, the operator determines the
best pH and chemical dosages by systematically changing
the level of the factor (i.e. pH, dosage) one step at a time
while holding the level of other factors constant. The level
of the factor that results in the best response (e.g. lowest turbidity value) is then selected and used in subsequent tests
which continue in the same manner for other factors.
While this approach, termed as one-factor-at-a-time
(OFAT), is rather straightforward, it suffers from shortcomings that may lead to a wrong conclusion. In particular, it
does not fully explore the whole experimental space to
nd the best factors conditions and is incapable of identifying the interaction effects resulting from the factors being
considered. Due to these reasons, the typical jar test practice
could have missed the actual best pH and dosage, which are
possibly hidden in the experimental space not covered by
the OFAT approach (Zainal-Abideen et al. ). Therefore,
a statistically designed experimental approach is proposed for jar tests to overcome the shortcomings formerly
stated.
Statistically designed experiments allow efciency and
are quite economical in that they require a relatively small
number of experiments but are still able to be analyzed by
statistical methods and result in valid and objective conclusions. The statistical approach to experimental design is
crucial if meaningful conclusions are to be drawn from the
data. When the problem involves data that are subjected
to experimental errors, a statistical method is the only systematic approach to analyse the data (Montgomery ).
The response surface method (RSM) is one example of the
statistical design of experiments. It is a collection of
mathematical and statistical techniques that are useful
for modelling and analyzing problems in which responses
(or a response) of interest are inuenced by several
factors or variables and in which the objective is to optimize
the responses. With RSM, the interaction between
factors and responses can also be determined (Montgomery
).
RSM has been used in many optimization experiments.
Ahmad et al. () explained the effect and interaction
between coagulant dose, occulant dose and pH when treating palm oil mill efuent (POME) through a coagulation
occulation process combined with membrane separation
technology. Wang et al. () optimized the coagulation

Water Science & Technology

65.3

2012

occulation process to achieve minimum turbidity and

sludge volume index (SVI) for paper-recycling wastewater
treatment. Pinzi et al. () used the RSM to optimize the
transesterication reaction for several types of vegetable
oils.
Despite the application of the RSM in many experimental studies, its use in optimizing coagulation conditions with
jar testing in water treatment is apparently lacking. Hence,
the objectives of the study were to obtain and to compare
the optimum coagulation conditions through traditional
jar testing conducted by Sri-Gading water treatment plant
(WTP) operators and the RSM. The variables that
were considered include the coagulants and coagulant aid
dosages, initial pH setting, nal turbidity, settling pH
after coagulation process and residual aluminium. As the
control of disinfection by-products is not a problem at
this WTP, NOM parameters were not included in this
investigation.

MATERIALS AND METHODS

Sri-Gading WTP
Sri-Gading WTP has a treatment capacity of 72,000 m3 of
water a day and supplies water for public use in
Batu-Pahat, Malaysia. Its water source is from the Bekok
River. For nearly a decade, the Bekok River has registered
a low pH of less than 5.5 due to the acidication
process of acid sulphate soil within the river basin (Aris
et al. ).
In identifying its best coagulation conditions, SriGading WTP has adopted an adjusted OFAT jar test
whereby only the coagulant dosing is systematically
altered and not the coagulant aid and pH of the coagulation. This situation is because the WTP is more
concerned about the amount of coagulant used as compared with any other chemicals as its usage is
considerably higher than the others. For the record, the
daily average cost ratio of coagulant aid to coagulant at
Sri-Gading WTP is 1:240. Apart from the coagulant aid,
the coagulation pH has also been kept constant during
the actual jar test as the WTP focuses more on the pH
after coagulation. As long as the pH after coagulation is
within 6.5 to 7.5 or around neutral, the initial pH for
coagulation is immaterial. Usually, this initial pH value
is set in the range of 7.28.0 and its mean value calculated
from the last years historical data is 7.55.

M. Zainal-Abideen et al.

498

Optimizing coagulation process through traditional & RSM jar test

Materials
Alum (liquid aluminium sulphate) and polymer (Nalclear
8173 PULV) respectively were used as the coagulant and
coagulant aid at Sri-Gading WTP. To adjust the raw water
pH to the desired pH, hydrated lime was used. Both alum
and hydrated lime used were obtained from Damini Corporation, Malaysia while polymer was delivered from Usaha
Kimia (Malaysia).

Water Science & Technology

65.3

2012

The optimum conditions for jar tests were selected when

the coagulated water from Sri-Gading WTP achieved
settling pH between 6.57.5, turbidity below 5 NTU and
residual aluminium less than 0.20 mg/l through minimum
usage of alum, hydrated lime and polymer. According to
World Health Organization guidelines (), the upper
limit of turbidity for nished water is 5 NTU and the
USEPA sets the secondary maximum contaminant level in
drinking water for aluminium to be 0.20 mg/l (Sawyer
et al. ).

Experimental setup
Traditional jar test
Raw water sample collected from Sri-Gading WTP was
brought to Universiti Teknologi Malaysia (UTM). This
sample was used to carry out traditional and statistical
designed (CCD RSM) jar test experiments at a bench
scale. Characterization of water quality was carried out in
accordance with Standard Methods (APHA et al. ).
Table 1 summarizes the water quality for the raw water
sample.
Before proceeding to the jar test experiment (conducted
using a Chemix Floc-Tester Model CL-4), the initial pH of
the water sample was adjusted by utilizing 1% hydrated
lime solution. Jar test mixing speeds and time were set according to the WTP practice: rapid mixing of one minute at
200 rpm; rapid mixing for 2 min at 100 rpm; slow mixing
for 3 min at 50 rpm; slow mixing of 10 min at 25 rpm; settling
time of 30 min. Alum, followed by polymer, was added at the
start of the fast (200 rpm) mixing stage. The ranges of initial
pH, alum and polymer concentration used in the jar tests
are explained below in the traditional jar test and RSM jar
test subsections.
Following the 30 min settling period, the pH of the solution was measured with a pH meter (model 410, Thermo
Orion) and samples were withdrawn 2 cm below the water
surface by using a 25 ml glass pipette (Hirschmann,
Germany). Samples were taken for turbidity and residual
aluminium analysis. Turbidity was determined by a turbidity
meter (HI 93703, Hanna Instrument). Residual aluminium
was measured through Aluminon method (adapted from
Standard Methods) using a spectrophotometer (DR 5000,
Hach Company, USA).

Table 1

Following the WTP practice, the alum dose was varied

between 818 mg/l for a water sample with turbidity
between 5 to 10 NTU and the initial pH for jar test was
raised to 7.6. It is also the WTP practice to set the polymer
dosage constantly at 0.02 mg/l. Table 2 shows the initial pH
and the dosages of alum and polymer used in these
experiments.
RSM jar test
A three-factor CCD with four replicates at the centre points
was employed in designing the RSM jar test. In total, 18 runs
were required in order to complete the experiment. This
approach is to t a quadratic surface which leads to optimization as well as experimental error for this study
(Multifactor RSM Tutorial ). Alum dosage (A), initial
pH (B) and polymer dosage (C ) used in this experiment
were the three independent variables in the coagulation process. The range and level of factors used are presented in
Table 3. They were developed from the pH and coagulants
dosing historical data used in the jar test conducted daily
at the plant for the last 1-year period (Zainal-Abideen
et al. ). The Design-Expert software (version 7.1, StatEase, Inc., Minneapolis, USA) was used for regression and
graphical analyses of the data. The optimum values for the
variables were obtained in two ways: (i) by direct reading
from RSM table (Table 4) and (ii) by carrying out the optimization procedure in the Design-Expert guide (Multifactor
RSM Tutorial ).

Water quality parameters for raw water sample from Sri-Gading WTP
W

pH

Turbidity (NTU)

Colour (TCU)

Fe (mg/l)

Al (mg/l)

Temperature ( C)

Mn (mg/l)

NH3 (mg/l)

TDS (mg/l)

4.04

5.46

30

0.52

1.75

22.6

0.07

0.15

87

M. Zainal-Abideen et al.

499

Table 2

Optimizing coagulation process through traditional & RSM jar test

Traditional jar test result

Table 4

Water Science & Technology

65.3

2012

CCD and response results for the study of three experimental variables in coded
units

Factors

Responses
Factors

Residual
Run

Responses

Alum

Initial

Polymer

Turbidity

Settling

aluminium

(mg/l)

pH

(mg/l)

(NTU)

pH

(mg/l)

Run

2

Turbidity

Settling

Residual

(NTU)

pH

aluminium (mg/l)

1.17

7.00

0.18

7.6

0.02

0.23

6.59

0.11

10

7.6

0.02

0.00

6.64

0.11

0.71

6.73

0.07

12

7.6

0.02

0.00

6.80

0.12

1

0.63

6.70

0.31

14

7.6

0.02

0.42

6.62

0.14

2

0.13

6.79

0.09

16

7.6

0.02

0.29

6.87

0.14

1

1.07

6.64

0.26

18

7.6

0.02

0.28

6.64

0.13

2

1.80

6.33

0.14

1.26

6.58

0.17

1.03

6.68

0.07

0.72

6.67

0.08

10

1

1

0.00

6.71

0.14

11

1.05

6.82

0.19

12

1

1

0.61

6.74

0.13

13

0.64

6.75

0.08

14

1.00

6.77

0.10

15

1

0.87

6.88

0.16

16

1

1

1

0.90

7.13

0.11

17

1

1

0.71

6.99

0.19

18

0.61

6.72

0.09

RESULTS AND DISCUSSION

Traditional jar test
Table 2 shows the results of the traditional jar test for the
water sample. It can be clearly observed that all runs gave
acceptable results as the responses are within acceptable
conditions. According to the WTP practice, the best condition for coagulation was chosen through direct reading
from Table 2 (i.e. run 3), where it gave the lowest turbidity
value, its settling pH is near to neutral (7) and the residual
aluminium is below 0.20 mg/l. Therefore, the chosen values
of alum dosage, initial pH and polymer dosage obtained
through traditional jar test were 12 mg/l, 7.6 and 0.02 mg/l
respectively.
RSM jar test
The role of the traditional jar test carried out at Sri-Gading
WTP was to obtain the optimum coagulation conditions
almost on the spot. If the RSM jar test is substituted for
the traditional jar test at the WTP, it is difcult for the operators to identify the optimum conditions immediately as the
results need to be analysed through the optimization procedure in the Design-Expert guide. Therefore, without
eliminating the necessity to execute the optimization procedure to obtain the best coagulation conditions, it is

Table 3

Experimental range and levels of central composite design

Range and levels

Variables

2

1

15.5

A, alum dosage (mg/l)

B, initial pH

6.5

6.9

7.5

24
8.1

30
8.5

C, polymer dosage (mg/l)

0.004

0.01

0.016

0.02

suggested that they look directly at the RSM results shown

in Table 4, as this table can identify one type of optimum
condition almost instantaneously.
From Table 4, it is evident that almost all runs gave an
acceptable quality of coagulated water except for runs 3, 5
and 6 in which the coagulated sample did not meet the
requirement in either settling pH or residual aluminium.
Through direct reading of the RSM table (Table 4), run 12
was decided to be the optimum coagulation condition as
its coagulated water had the third lowest turbidity and satised other requirements. Although runs 4 and 10 produced
even lower turbidity coagulated water than run 12 and
met settling pH and residual aluminium conditions, they utilized higher alum dosages. The conditions for run 12 were
alum dosage: 7 mg/l, initial pH: 6.9 and polymer dosage:
0.016 mg/l.
In order to obtain the optimum coagulation conditions
through the optimization procedure in the Design-Expert
guide, the regression models formulated by the software
relating responses and factors in terms of coded factors
need to be examined. Whenever necessary, statistically
insignicant terms were eliminated through backward
method and/or response transformation were carried out

M. Zainal-Abideen et al.

500

Table 5

Optimizing coagulation process through traditional & RSM jar test

Water Science & Technology

65.3

2012

ANOVA results for response parameters in RSM jar test

Model
F value

Response

Prob > F

LOF
Prob > F

R2

Adjusted R 2

AP

SD

CV

PRESS

Turbidity

3.82

0.0455

0.1164

0.3376

0.2493

6.034

0.35

41.66

2.44

Settling pH

5.63

0.0096

0.0050

0.5469

0.4499

8.040

0.13

1.92

0.44

Residual aluminium

6.03

0.0059

0.2995

0.8085

0.6744

6.696

0.35

12.30

5.05

LOF: lack of t; AP: adequate precision; SD: standard deviation; CV: coefcient of variation; PRESS: prediction error sum of squares.

to form signicant models. The generated multiple

regression equations are as follows:
Turbidity 1:00 0:14C
0:23C 2

Settling pH 6:76
0:099A 0:063B
0:084C

1
p 3:42 0:084A
0:050B 0:012C 0:42AB
Residual Al
0:22BC
0:40A2
0:36B2

The analysis of variance (ANOVA) for the regression

models (1), (2) and (3) are summarized in Table 5. A
model is signicant at the 95% condence level if the
Fisher F-test has a probability value (Prob > F ) below 0.05.
The lack of t (LOF) F-test describes the deviation of
actual points from the tted surface, relative to pure error
(Anderson & Whitcomb ). A large value of Prob > F
for LOF, possibly greater than 0.05, is preferred. A high R 2
value is desirable and a reasonable agreement with adjusted
R 2 is crucial (Ghafari et al. ). Adequate precision (AP) is
dened as a measure of the experimental signal to noise
ratio (Anderson & Whitcomb ); an AP that exceeds 4
usually indicates that the model will give reasonable performance in prediction. The standard deviation (SD) and
the coefcient of variation (CV) are shown in Table 5.
PRESS, the prediction error sum of squares, is a measure
on how well the model for the experiment is likely to predict
the responses in a new experiment. The SD, CV and PRESS
values are preferred to be small (Montgomery ).
The regression model (1) is a signicant model as its
Prob > F value is smaller than 0.05. In this particular
water sample, the second-order effect of polymer (C 2) is
the only signicant model term. The function of polymer
is to improve the effectiveness of coagulant in coagulation
process. Long-chain charged synthetic and natural polymers
(polyelectrolytes) can act to destabilize colloids by forming a
bridge between one colloid and another. One charge site on

the long polymer can adsorb onto a site on one colloid,

while the remainder of the polymer molecules extend into
solution. If the extended portion of the polymer becomes
attached to another colloid, then the two colloids are effectively tied together and form ocs (Sawyer et al. ). The
ocs may then be easily removed through the occulation
process; hence reducing the turbidity of water.
The Prob > F for LOF is large and this situation implies
that the LOF of the model is insignicant. Even though the
R 2 value is relatively low, it is in good agreement with
adjusted R 2. It is worth pointing out that a good regression
model does not depend on a large value of R 2 as R 2 is not
very well suited to assess outcomes from planned experiment (Myers & Montgomery ; Anderson &
Whitcomb ). The AP, SD and PRESS for the regression
model (1) are satisfactory as the AP is greater than 4 and SD
as well as PRESS values are very small.
The regression model (2) is also a signicant model with
a Prob > F value of 0.0096. The main effect of alum (A) and
polymer (C ) dosage are the signicant model terms. From
the regression model, the addition of alum will reduce
the pH. This may be explained by Al (III) ions hydrolysis
phenomenon to form soluble monomeric (Equation (4)) as
well as polymeric species, H ion and solid precipitates
when alum is mixed with water (Jiang & Graham ).
This will reduce the pH of treated water.
Al3 H2 O ! AlOH2 H

The regression model (2) has a relatively good R 2 value

(0.5469) and is in good agreement with the adjusted R 2. The
AP value was greater than 4 while the SD, CV and PRESS
values are small.
A signicant model with Prob > F value of 0.0059 is produced to form regression model (3) when the response has
been transformed by inverse square root to provide a
better t. In this case, the two level interactions of alum
dosage and initial pH (AB) as well as the second-order
effect of alum dosage (A 2) and initial pH (B 2) are the

501

Figure 1

M. Zainal-Abideen et al.

Optimizing coagulation process through traditional & RSM jar test

Design-expert plot; predicted vs. actual values plot for (a) turbidity, (b) settling pH, (c) 1=

signicant model terms. The Prob > F value of LOF is 0.2995

and this nding indicates that the LOF is not signicant relative to the pure error. The R 2 value is high (0.8085) and does
not deviate far from the adjusted R 2 value. The AP, SD, CV
and PRESS values comply with the statistical conditions
listed previously. Diagnostic plots such as the predicted
versus actual values (Figure 1) help us to judge if the
model is satisfactory. These plots indicate an adequate agreement between real data and the outputs from the models.
Desirable criteria were set up to obtain the optimum jar
test condition through RSM. The criteria were minimum
dosages of alum and polymer; initial pH setting and settling
pH after coagulation were in the ranges of 6.98.1 and 6.5
7.5 respectively; turbidity must be minimum and below
5 NTU and residual aluminium must be below 0.20 mg/l.
With multiple responses, the optimum condition is one at
which all parameters simultaneously meet the said desirable
criteria. This result could be visualized graphically by superimposing the contours of the response surfaces of the
regression models (1), (2) and (3) in an overlay plot. Graphical optimization displays the area of feasible response values
in the factor space and the regions that do t the optimization criteria would be shaded (Ghafari et al. ). The
shaded area in Figure 2 shows the RSM optimum jar test
condition for the desirable criteria mentioned earlier.
One of the conditions in the region is displayed in
Table 6. Experiments were conducted to determine whether
the predicted results by the models are attainable. The experiments were carried out in triplicate and the responses
expressed in Table 6 were the mean of the three experimental results. From these results, it is observed that the
responses predicted through the regression models and
measured from the experiments were in close agreement.

Water Science & Technology

65.3

2012

p
Residual aluminium of coagulated water sample through RSM jar test.

Figure 2

Design-Expert plot; overlay plot for optimal region at polymer 0.004 mg/l.

Traditional vs. RSM jar test

The alum dose for optimum coagulation obtained through
the traditional method was nearly double the one obtained
through the RSM jar test. For the polymer, its dosing in
the traditional jar test has always been set to be 0.02 mg/l
regardless of the condition of the raw water. On the contrary, for this particular raw water sample, it was observed
from the RSM experiment that it was not necessary to use
0.02 mg/l of polymer. Instead, the optimum polymer doses
obtained from direct reading of RSM table and RSM optimization were 0.016 and 0.004 mg/l, respectively.
Nonetheless, the initial pH set for coagulation in the traditional jar test that produced coagulated water that satised
the WTP requirement was the same as the initial pH setting
for optimum coagulation acquired through RSM optimization. The pH setting obtained through direct reading of
RSM table was amongst the lowest. Although the water turbidity after applying the traditional method was lower than

M. Zainal-Abideen et al.

502

Table 6

Optimizing coagulation process through traditional & RSM jar test

Experimental and predicted values of the responses at the optimal levels

predicted by RSM

Optimal conditions

Response

Predicted

Measureda

A 7 mg/l; B 7.6;
C 0.004 mg/l

Turbidity
(NTU)

0.64

0.83 0.20

Settling pH

7.00

7.21 0.02

Residual
aluminium (mg/l)

0.13

0.15 0.04

The measured values are expressed as means standard deviations of three indepen-

dent experiments.

the RSM jar test, both methods gave water turbidity below
1 NTU. The pH values after ocs settlement acquired
through both methods were lower than their initial pH settings and met the optimum condition. The residual
aluminium concentrations from both traditional and RSM
jar tests were below 0.20 mg/l.

CONCLUSIONS
The quest for optimum coagulation operational pH, alum
and polymer dosages through RSM has been successful. It
proved that the RSM jar test generated lower optimum
alum and polymer dosages than the traditional jar test and
was still able to produce comparable and acceptable quality
of coagulated water in terms of pH after ocs settlement, turbidity and residual aluminium. The interaction between
these factors and responses in coagulation process were
clearly demonstrated through RSM.
RSM jar testing requires a greater number of runs compared with the adjusted OFAT or traditional jar test. Should
the actual OFAT jar test is be implemented at Sri-Gading
WTP, similar number of runs may have to be conducted as
the polymer dosing and the initial pH for coagulation
must also be varied. Nevertheless, the purpose of this investigation is not to replace the traditional jar test by RSM but is
a part of an investigation to form an empirical relationship
between raw water quality parameters and the best coagulation conditions obtained through traditional and RSM
methods in term of pH and chemical dosing.

ACKNOWLEDGEMENTS
The authors thank Syarikat Air Johor (Holdings) Sdn. Bhd.
for providing nancial support and giving permission to
work with Sri-Gading WTP in order to carry out this study
successfully.

Water Science & Technology

65.3

2012

REFERENCES
Ahmad, A. L., Ismail, S. & Bhatia, S. Optimization of
coagulationocculation process for palm oil mill efuent
using response surface methodology. Environmental Science &
Technology 39 (8), 28282834.
Amirtharajah, A. & OMelia, C. R. Coagulation processes:
destabilization, mixing, and occulation. In: Water Quality
and Treatment A Handbook of Community Water Supplies
(F. W. Pontius, ed.). 4th edition, McGraw-Hill, Inc.,
New York, USA, pp. 269366.
Anderson, M. J. & Whitcomb, P. J. RSM Simplied.
Productivity Press, New York.
APHA/AWWA/WEF Standard Methods for the Examination
of Water and Wastewater, 21st edition, American Public
Health Association/American Water Works Association/
Water Environment Federation, Washington DC, United
States of America.
Aris, A., Mokhtar, N., Ujang, Z. & A-Majid, Z. Treatment
of acidic water from Bekok River using limestone channel.
Water Science & Technology: Water Supply 7 (4), 6571.
Brub, D. & Dorea, C. C. Optimizing alum coagulation for
turbidity, organics, and residual Al reductions. Water Science
& Technology: Water Supply 8 (5), 505511.
Ghafari, S., Abdul Aziz, H., Isa, M. H. & Zinatizadeh, A.
Application of response surface methodology (RSM) to
optimize coagulationocculation treatment of leachate
using poly-aluminum chloride (PAC) and alum. Journal of
Hazardous Materials 163 (23), 650656.
Gregory, D. & Carlson, K. Relationship of pH and oc
formation kinetics to granular media ltration
performance. Environmental Science & Technology 37 (7),
13981403.
Hering, J. G., Chen, P., Wilkie, J. A. & Elimelech, M. Arsenic
removal from drinking water during coagulation. Journal of
Environmental Engineering 123 (8), 800807.
Jiang, J. & Graham, N. Pre-polymerised inorganic coagulants
and phosphorus removal by coagulation a review. Water SA
24 (3), 237244.
Lin, J., Huang, C., Pan, J. R. & Wang, D. Effect of Al(III)
speciation on coagulation of highly turbid water.
Chemosphere 72 (2), 189196.
Montgomery, D. C. Design and Analysis of Experiments.
John Wiley and Sons, Inc., United States of America.
Morfesis, A., Jacobson, A. M., Frollini, R., Helgeson, M., Billica, J. &
Gertig, K. R. Role of zeta () potential in the optimization
of water treatment facility operations. Industrial &
Engineering Chemistry Research 48 (5), 23052309.
Multifactor RSM Tutorial (Part 1 The Basics) Design-Expert
7.1 Users Guide. Stat-Ease, Inc., Minneapolis.
Myers, R. H. & Montgomery, D. C. Response Surface
Methodology: Process and Product Optimization Using
Design Experiments. Wiley & Sons, USA.
Pinzi, S., Lopez-Gimenez, F. J., Ruiz, J. J. & Dorado, M. P.
Response surface modeling to predict biodiesel yield in a
multi-feedstock biodiesel production plant. Bioresource
Technology 101 (24), 95879593.

503

M. Zainal-Abideen et al.

Optimizing coagulation process through traditional & RSM jar test

Sawyer, C. N., McCarty, P. L. & Parkin, G. F. Chemistry For

Environmental Engineering. McGraw-Hill, Singapore.
Wang, J., Chen, Y., Ge, X. & Yu, H. Optimization of
coagulationocculation process for a paper-recycling
wastewater treatment using response surface methodology.
Colloids and Surfaces A: Physicochemical and Engineering
Aspects 302 (13), 204210.
WHO Guidelines for Drinking-water Quality, 3rd edition.
World Health Organization, Geneva.

Water Science & Technology

65.3

2012

Zainal-Abideen, M., Aris, A., Yusof, F. & Abdul-Majid, Z.

Optimization of Coagulation Conditions Using Statistical
Approach. Progress Report to Syarikat Air Johor Holdings
Sdn. Bhd., Johor Bahru, Malaysia.
Zhan, X., Gao, B., Yue, Q., Wang, Y. & Cao, B. Coagulation
behavior of polyferric chloride for removing NOM from
surface water with low concentration of organic matter and
its effect on chlorine decay model. Separation and
Purication Technology 75 (1), 6168.

First received 8 November 2010; accepted in revised form 18 February 2011