Desalination 277 (2011) 74–82

Contents lists available at ScienceDirect

Desalination
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / d e s a l

Optimization of coagulation and dissolved air flotation (DAF) treatment of

semi-aerobic landfill leachate using response surface methodology (RSM)
Mohd Nordin Adlan ⁎, Puganeshwary Palaniandy, Hamidi Abdul Aziz
School of Civil Engineering Campus, Universiti Sains Malaysia, 14300 Nibong Tebal, Pulau Pinang, Malaysia

a r t i c l e i n f o a b s t r a c t

Article history: FeCl3 coagulation and dissolved air flotation (DAF) were combined to assess the success of these techniques
Received 14 December 2010 for the treatment of semiaerobic landfill leachate. Treatment parameters (i.e.; flow rate, coagulant dosage, pH
Received in revised form 30 March 2011 and injection time) were optimized via response surface methodology (RSM) using central composite design
Accepted 1 April 2011
(CCD) to yield the maximum removal of turbidity, chemical oxygen demand (COD), color and ammonia
Available online 4 May 2011
nitrogen (NH3-N). Model-determined optimum conditions were tested to confirm the predicted results.
Keywords:
Initial concentrations of turbidity (259 FAU), COD (2010 mg/L), color (4000 PtCo) and NH3-N (1975 mg/L)
Dissolved air flotation were reduced by 50%, 75%, 93% and 41%, respectively. These experimental results were consistent with those
Coagulation predicted by the model. The optimum operating conditions for coagulation and DAF were 599.22 mg/L of
Ferric chloride FeCl3 at pH 4.76 followed with saturator pressure of 600 kPa, flow rate 6 L/min and injection time of 101 s.
Landfill leachate © 2011 Elsevier B.V. All rights reserved.
Response surface methodology

1. Introduction In addition to the successful removal of contaminants and

production of thick sludge, DAF provides further benefits for wastewater
Landfill dumping is one of the most abundant methods in developing treatment, including increases in dissolved oxygen and decreases in
countries for municipal solid waste disposal; however, this method is color, algae, turbidity, and oil and grease [15,16]. DAF treatment also
considerably hazardous to humans and the environment due to the offers low capital investment, low operational cost and high separation
leaching of gasses and contaminated wastewater. The latter is efficiency [17,18].
concentrated with organic and inorganic contaminants that can be To successfully perform DAF, a coagulation process must be
either biodegradable or refractory in the environment, including humic introduced to facilitate the destabilization of colloidal particles or
substances, ammonia nitrogen, heavy metals, and chlorinated organic emulsions. This is generally accomplished using four different mecha-
and inorganic salts [1–4]. These compounds are the main contributors to nisms, including double layer compression, charge neutralization,
high COD, nitrogen, turbidity and color pollution [5]. entrapment in a precipitate and inter-particle bridging [19].
Landfill leachate requires a series of biological, chemical, and The charge of the particle and bubble is the most important factor
physical treatments that are complicated and expensive [6–9]. for inducing flotation and removing pollutants, [20].
Biological treatments involving aerobic and acetogenic degradation Previously, DAF treatment has successfully removed 60% of humic
are effective for removing relatively young, recently produced leachate, acids (which contribute to COD and color concentrations in leachate)
such as volatile fatty acids [10]. Older leachate contains more from synthetic landfill leachate [21]. The aim of this study was to
biorefractory contaminants and additional pollutants (i.e. ammonia) optimize the combination of coagulation and DAF processes for actual
that are produced during acetogenic and methanogenic phases of landfill leachate treatment using RSM. This study examined the
degradation. In order to cope with the changing chemical composition individual and interactive influences of coagulation and flotation factors
of the leachate, treatment processes combining additional chemical and (i.e., pressure, flow rate, coagulant dose, pH, and bubble injection time)
physical mechanisms have been applied [3,5,11,12]. For example, the on the removal of turbidity, color, COD, and NH3-N from landfill leachate.
dissolved air flotation (DAF) treatment method which relies on the
collision of air bubbles with the suspended particles could float the latter 2. Methods and materials
to the surface of the flotation tank and thus help to reduce leachate
contamination [13,14]. 2.1. Sampling and characterization

Raw leachate was collected from Pulau Burung sanitary landfill,

located within the Byram Forest Reserve in Penang, Malaysia (5° 24′ N,
⁎ Corresponding author. Tel.:+60 4 599 6252; fax: +60 4 594 1009. 100° 24′E). The landfill was developed in 1980's by Seberang Prai
E-mail address: cenordin@eng.usm.my (M.N. Adlan). Municipal Council. The pollutant composition of the waste stream at

0011-9164/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.desal.2011.04.006
 M.N. Adlan et al. / Desalination 277 (2011) 74–82 75

Table 1 and was converted into a semiaerobic Level 2 sanitary landfill by

Mean waste composition of municipal solid waste in Pulau Burung Sanitary Landfill, establishing a controlled tipping technique in 1991 [1]. In 2001 this
[22].
landfill was further upgraded to Level 3 by establishing controlled
Waste Composition Waste Composition tipping with leachate recirculation. The piping system collects leachate
components components at the bottom of the landfill and channels it into a collection pond [23].
(% weight) (% weight)
The capacity of the collection and retention ponds are 200 m3 and
Food waste 40.08 Textile 5.08
Paper 12.89 Yard 9.23 11400 m3, respectively. Raw leachate used in this study was sampled
Plastics 24.04 Glass 3.28 from the collection pond; the characteristics of the leachate are shown
Wood 1.54 Ferrous 0.47 in Table 2.
Rubber 1.29 Aluminum 2.10 Leachate samples were collected in 2 week intervals from April
2008 to September 2008. The sampling method was based on the
Standard Methods for the Examination of the Water and Wastewater,
Table 2
[24]. The pH and dissolved oxygen (DO) of leachate samples were
Characterization of Pulau Burung landfill leachate.
analyzed on site, while COD, color, turbidity, NH3-N, alkalinity, and
Parameter Maxa Mina Averageb suspended solid (SS) measurements were performed immediately
pH 8.24 7.84 8.13 upon return to the laboratory [24].
COD (mg/L) 3041 2370 2610
Color (PtCo Unit) 4932 3625 4000
Turbidity (FAU) 386 207 259
Alkalinity (mg/L as CaCO3) 11,000 8117 10,416
2.2. Materials and chemicals
Suspended solids (mg/L) 317 177 218
Dissolved oxygen (mg/L) 4.10 0.09 0.58 The bench scale DAF unit used for this study was designed to
Ammoniacal nitrogen (mg/L) 2117 1797 1975 perform coagulation and DAF in the same cell. It consists of an
a
The values are the average of 3 replicates. Replicate analyses varied by less than 1%. unpacked saturator, a flotation cell, a compressor, a water tank and a
high pressure pump. The air injection nozzle in the flotation cell (USM
b
An average of 10 samples were collected at 2 week intervals from April 2008 to
September 2008.
nozzle) was developed by Adlan and co-workers [25]. The sampling
point is located 7.5 cm above the floor of the flotation cell and a scour
valve was installed at the bottom of the flotation cell (Fig. 1).
Pulau Burung Sanitary Landfill is presented in Table 1 [22]. The relative Ferric chloride (FeCl3; supplied by R&M Chemicals, UK) was
humidity in this area ranges between 70 and 90% and the annual chosen to induce coagulation. Sodium hydroxide (NaOH) and sulfuric
average rainfall is 2670 mm. The landfill area is approximately 23.7 ha acid (H2SO4) were used to adjust pH during the coagulation process.

Fig. 1. Schematic diagram of the DAF batch study.

76 M.N. Adlan et al. / Desalination 277 (2011) 74–82

2.3. Experimental set-up and procedure preliminary experiments [27]. Another factor that was concern in this
study is the injection time. The maximum value for the injection time was
The coagulation and DAF processes were performed as follows: set at 120 s in order to reduce the dilution effect in the batch study. The
(i) the pH of the leachate (4 L) was adjusted according to the experiment injection times were varied from 30 up to 120 s. However, increasing of
and the sample was added to the flotation cell, (ii) FeCl3 was added to the injection time will also increase the number of bubbles in the flotation
cell according to the design of experiment (iii) the leachate and FeCl3 cell, which simultaneously promotes more bubble particle attachment
were rapidly mixed (470 rpm for 5 min), (iv) water saturated with air and thus more particles will be floated. The values of pressure and flow
was injected from the saturator into the flotation cell for 2 min, and rate were set based on the research work conducted by Edzwald et al. [28]
(v) flotation was allowed to occur for 20 min and samples were collected in order to produce microbubbles in the flotation cell. This has been
from the sampling point. Treated leachate samples were analyzed using proved by Al-Shamrani et al. [14] in her study on treating oily wastewater.
the standard method [24]. COD was measured according to Method Four main parameters were chosen to evaluate the effectiveness of
5220D [24], (closed reflux, colorimetric method). Colors were reported the coagulation/DAF process, including turbidity (Y1), COD (Y2), color
as true color (filtered using 0.45 μm filter paper) and were determined (Y3) and NH3-N (Y4).
using a DR 2010 HACH spectrophotometer, as in Method 2120C [24]. Generally, the CCD consists of a 2k factorial with nF factorial runs
Turbidity was determined using a DR2010 HACH spectrophotometer. (points with all possible combination of the minimum and maximum
NH3-N was measured according to the Nessler method and using a DR values of the control parameters), 2k axial or star runs (one of the
2010 HACH spectrophotometer as in Method 4500C. Alkalinities were parameters has the minimum or maximum value and all other
measured using 2320B titration methods [24] and are reported as mg/L parameters have their nominal value), and nC center runs (all control
of calcium carbonate. SS were determined using a DR 2010 HACH parameters are set to their nominal values). In this study, a total of 50
spectrophotometer, similar to Method 2540D [24]. The pH and DO were experiments were performed to assess the five experimental factors,
measured using a W-100 Witeg pH meter and WTW multi-parameter according to the equation CCD = 2k + 2 k + 8, where k is the number
340i, respectively. of factors. Forty-two experiments were improved with eight replica-
tions at the design center to evaluate the pure error, [26]. Eq. (2)
2.4. Calculations shows the quadratic model used to estimate the optimal point:

The removal of the studied parameters from leachate was k k k k

2
calculated based on the following formula: Y = β0 + ∑ βi :Xi + ∑ βii :Xi + ∑ ∑ βij :Xi :Xj + … + e ð2Þ
i=1 i=1 ii≤j j
 
Co −Ci
Removal Percentage = × 100 ð1Þ
Co where Y is the response; Xi and Xj are the variables; β0 is a constant
coefficient; βi, βii, and βij are the interaction coefficients of linear, quadratic
where Co and Ci are the initial and final concentrations of the studied and second-order terms, respectively; k is the number of studied factors;
parameters. Here the concentration was based on absolute contents in and e is the random error.
the flotation cell (including the dilution effect of the tap water The coefficient of determination (R2) was used to identify the
introduced for the flotation) [9]. quality of the fit of polynomial model and the P-value associated with
the 95% confidence level was used to evaluate the variables and the
2.5. Experimental design, analysis and optimization interactions between them. The significance and adequacy of the
model was assessed according to the calculated F-value (Fisher
The design, mathematical modeling and optimization of this study variation ratio), probability value (ProbNF), and Adequate Precision.
were performed using Design Expert 6.0.7 software. Central composite Finally, the values of response variables (flow rates, pressures,
design (CCD) was used to model the RSM in this design. The former is the dosages, injection times and pH) were compared in an overlay plot
most widely used experimental design for fitting a second-order response to identify the optimum region of leachate treatment.
surface [26]. The independent variables (factors) used in this experimen-
tal study were pressure, flow rate, dosage of coagulant, pH and injection 3. Results and discussion
time, and are coded as A, B, C, D, and E respectively (Table 3). The
independent variables were varied over three levels, between −1, 0 and The leachate used in this study, exhibited low BOD5/COD ratio
+1, and the range was determined based on preliminary studies and (0.03–0.06), high alkalinity and high NH3-N, which can be classified as
literature review. In the preliminary study, the responses became “stabilized leachate” and resistant to biodegradation, [29].
constant after the dosage concentration was increased up to 1500 mg/L. Leachate contains high concentration of bacteria, organic substances
For the pH, FeCl3 worked well in the acidic range, this was also found by and suspended particles. Generally, particles carry a negative charge on
Aziz et al. [7]. Mixing intensity at 470 rpm (5 min) and flotation time the surface of the liquid in which the pH ranges from 5 to 9, with a typical
(20 min) were kept constant for each experiment; these factors appeared size range of 0.01–1 μm [30–32]. Coagulation increases the size and
to be insignificant for coagulation and flotation as determined in the alters the surface characteristics of the particles. In a leachate solution
adjusted to be slightly acidic (pH 4 to 6), ferric chloride disassociates into
Fe3+, Fe(OH)+ 2 and FeOH
2+
ions. In this phase, charged particles are
Table 3
neutralized by the coagulant. This results in the formation of flocs, which
Actual and coded values for the independent variables of the CCD design. lead to more bubble-particle attachment during the DAF process. The
bubble-particle attachment involves three mechanisms; (i) precipitation
Factor (symbol) Coded value
or collisions, (ii) bubbles trapped in a floc structure as the bubbles rise
−1 0 +1 through the liquid media and (iii) bubbles adsorbed in a floc structure as
Actual value the floc is formed, [21]
A. Pressure (kPa) 400 500 600 In this experiment, a total of 50 CCD batch runs were conducted
B. Flow rate (B) 4 5 6 (Table 4). Using these treatments, the highest percentage removal for
C. Coagulant dose (mg/L) 125 812.5 1500 turbidity was 54% (Table 4). However, high dosages of FeCl3 increased
D. pH 4 5 6 turbidity due to the formation of heavy sludge that did not float to the
E. Injection Time (s) 30 75 120
surface of the flotation cell; this observation is consistent with
 M.N. Adlan et al. / Desalination 277 (2011) 74–82 77

Table 4
Results of the central composite design.

Run Process variable Responses

A B C D E Actual Model predicted

1 2 3 4 1 2 3 4

1 500 5 812.5 5 75 43.2 71.6 91.7 37.5 2.74 32.37 50.21 10.88
2 600 4 125 4 30 − 4.8 36.5 54.9 9.5 2.74 36.70 55.62 10.88
3 500 5 812.5 5 120 54.0 74.6 95.1 39.3 2.74 30.20 50.21 10.88
4 600 4 125 6 30 − 6.4 28.7 34.9 3.0 2.74 34.53 55.62 10.88
5 600 4 1500 6 30 −80.8 50.5 78.5 6.7 − 31.38 53.44 65.84 19.43
6 400 4 125 4 30 −12.1 32.3 51.6 8.0 − 31.38 57.77 71.26 19.43
7 400 6 125 4 30 − 4.0 30.7 51.9 11.9 − 31.38 55.61 65.84 19.43
8 400 5 812.5 5 75 29.9 68.8 93.8 33.5 − 31.38 59.94 71.26 19.43
9 500 6 812.5 5 75 7.8 70.8 90.8 38.1 − 24.21 26.46 35.65 10.88
10 600 6 125 6 30 − 0.3 32.0 37.8 2.8 − 24.21 30.79 41.07 10.88
11 600 4 1500 4 120 −37.2 74.3 89.6 49.5 − 24.21 24.29 35.65 10.88
12 600 4 125 6 120 26.7 57.9 60.8 40.0 − 24.21 28.62 41.07 10.88
13 500 5 812.5 5 30 46.7 64.4 94.1 9.3 − 58.33 47.52 71.25 19.43
14 600 6 125 4 30 0.4 33.4 53.6 16.6 − 58.33 51.85 76.66 19.43
15 500 5 812.5 5 75 45.2 72.3 95.7 27.1 − 58.33 49.69 71.25 19.43
16 600 5 812.5 5 75 16.6 77.5 93.8 41.5 − 58.33 54.02 76.66 19.43
17 500 5 1500 5 75 15.5 71.0 89.4 38.3 32.63 55.89 64.98 40.54
18 600 4 125 4 120 30.1 59.4 74.6 48.6 32.63 60.22 70.40 40.54
19 400 6 1500 6 30 −48.5 54.4 80.6 18.2 32.63 53.72 64.98 40.54
20 500 5 125 5 75 15.8 46.9 56.5 29.7 32.63 58.05 70.40 40.54
21 500 5 812.5 5 75 44.9 72.6 91.8 34.6 − 1.48 67.97 80.62 49.08
22 600 6 1500 6 30 − 104.6 53.5 80.1 30.6 − 1.48 72.30 86.03 49.08
23 500 5 812.5 6 75 − 48.3 53.3 70.0 36.1 − 1.48 70.14 80.62 49.08
24 500 5 812.5 5 75 46.5 72.8 94.9 31.0 − 1.48 74.47 86.03 49.08
25 400 4 1500 6 120 − 63.1 63.8 79.4 47.9 5.68 49.98 50.43 40.54
26 400 4 125 6 30 − 20.6 27.7 55.6 12.2 5.68 54.31 55.84 40.54
27 500 4 812.5 5 75 48.8 72.2 89.0 36.9 5.68 47.81 50.43 40.54
28 500 5 812.5 4 75 − 11.5 65.5 80.9 39.2 5.68 52.14 55.84 40.54
29 400 6 125 4 120 26.4 54.3 64.8 42.6 − 28.44 62.06 86.03 49.08
30 400 4 1500 4 30 − 52.8 57.7 53.4 28.1 − 28.44 66.39 91.44 49.08
31 400 6 125 6 30 − 12.5 22.9 19.6 11.9 − 28.44 64.23 86.03 49.08
32 400 4 125 6 120 19.1 51.2 54.4 41.2 − 28.44 68.56 91.44 49.08
33 600 6 125 6 120 39.3 53.7 48.7 49.8 35.85 68.94 88.21 35.53
34 400 4 125 4 120 30.2 54.2 64.0 42.8 35.85 73.27 93.63 35.53
35 400 6 1500 4 120 −11.7 69.7 78.9 41.4 35.85 71.11 90.92 35.53
36 400 6 125 6 120 −41.2 49.2 59.8 41.5 35.85 71.11 90.92 35.53
37 600 4 1500 4 30 8.7 59.6 82.1 23.2 52.91 51.76 64.33 31.26
38 500 5 812.5 5 75 35.3 70.5 86.6 43.2 18.79 70.51 89.95 39.80
39 600 6 1500 6 120 43.6 73.5 93.2 50.8 0.63 64.56 81.90 35.53
40 400 4 1500 6 30 −80.4 41.8 66.3 13.9 − 26.32 58.64 77.33 35.53
41 400 4 1500 4 120 − 4.2 68.9 81.3 55.0 20.90 61.60 83.53 15.16
42 600 6 125 4 120 30.8 56.8 69.4 34.6 50.79 80.62 98.31 44.81
43 600 4 1500 6 120 − 21.2 61.2 89.7 46.6 35.85 71.11 90.92 35.53
44 500 5 812.5 5 75 50.3 71.8 89.6 33.9 35.85 71.11 90.92 35.53
45 600 6 1500 4 30 − 27.3 54.9 66.5 29.8 35.85 71.11 90.92 35.53
46 400 6 1500 6 120 − 48.5 61.8 84.9 42.4 35.85 71.11 90.92 35.53
47 500 5 812.5 5 75 25.2 70.9 86.5 30.6 35.85 71.11 90.92 35.53
48 500 5 812.5 5 75 48.0 69.2 88.0 37.5 35.85 71.11 90.92 35.53
49 400 6 1500 4 30 6.8 55.0 65.1 22.0 35.85 71.11 90.92 35.53
50 600 6 1500 4 120 42.8 74.7 89.3 47.8 35.85 71.11 90.92 35.53

A, B, C, D, and E = pressure, flow rate, dosage, pH and injection time.

1, 2, 3, and 4 = turbidity, COD, color and ammonia nitrogen.

previous reports [14]. COD was reduced by 22.9 to 77.5%, color by 19.6 concentration of 1975 mg/L, was achieved at pH 4, 400 kPa, 4 L/min flow
to 95.7%, and NH3-N by 2.8 to 55% (Table 4). COD was reduced by a rate, and 120 s injection time, and FeCl3 dose of 1500 mg/L.
maximum of 78%, from an initial concentration of 2610 mg/L, using
the following treatment conditions: pressure, 600 kPa; flow rate, 5 L/ 3.1. Regression model equation and analysis of variance (ANOVA)
min; FeCl3 dose, 812.5 mg/L; pH 5; and injection time, 75 s. Using a
coagulation/flocculation treatment, Tatsi and co-workers [33] dem- Based on the sequential model sum of squares, the models for
onstrated a reduction in COD by 80% with a FeCl3 dosage of 1500 mg/L turbidity, COD, color and NH3-N percentages removal were selected based
and flotation time of 20–30 min [7]. The coagulation/DAF treatment on the highest order polynomials where the additional terms were
used in the current study required only 5–10 min and also reduced significant and the models were not aliased. The models were coded as Y1,
the amount of sludge produced after treatment. Y2, Y3 and Y4 for turbidity, COD, color and NH3-N, respectively.
Reduction in color was considerably high, with the highest removal of The quadratic model for all four terms, Y1, Y2, Y3 and Y4, were selected
96%, from an initial measurement of 4000 PtCo units. This was achieved as suggested by the software and are shown in Eqs. (3)–(6). The
using 812.5 mg/L FeCl3 with pH 5, at 500 kPa, 5 L/min flow rate and 75 s independent variables in the models were pressure, flow rate, dosage, pH
injection time. The maximum removal of NH3-N (55%), from an initial and injection time and were coded as A, B, C, D, and E respectively. The
78 M.N. Adlan et al. / Desalination 277 (2011) 74–82

final empirical models used to generate coded factors for each variable determination (R2) for each empirical equation from Eq. (3)–Eq. (6),
are as follows: were 0.65, 0.97, 0.89 and 0.86 respectively. Three out of four models
(Y2, Y3 and Y4) show a good agreement between the experimental and
2
Y1 = 35:85−17:06 C−13:48D + 14:95E−48:70D ð3Þ model-predicted values. The standard deviations for the models were
24.58, 2.89, 6.50, and 5.44 for Y1, Y2, Y3 and Y4 respectively. Here, for
2 2 turbidity removal, the coefficient of determination was considerably
Y2 = 71:11 + 2:16A + 9:37C−2:96D + 9:51E–9:97C –9:51D ð4Þ
low and the standard deviation model was fairly high compared to
+ 1:09BC–2:25CE
other models. As shown in Table 5, 65%, 97%, 89% and 86% of the total
2 2
variability in turbidity, COD, color and NH3-N removal percentages
Y3 = 90:92 + 2:71A + 12:81C−2:29D + 7:39E−13:78C −11:31D ð5Þ were accredited to the empirical model, respectively. The ANOVA
+ 4:99CD (Table 5) revealed that all independent variables were significant
(p b 0.05) for determining turbidity, COD, color, and NH3-N.
2
Y4 = 35:53 + 4:27 C + 14:83E–5:55E : ð6Þ
3.2. Effect of factors on turbidity, COD, color, and NH3-N removal
The quality of the model was evaluated based on the coefficient of
determination in addition to the ANOVA statistical analysis. The Perturbation plots were analyzed in order to further identify the
ANOVA results for the quadratic model for turbidity, COD, color and most sensitive factors for leachate treatment (Fig. 2). Coagulant dose,
NH3-N percentage removal are shown in Table 5. The coefficients of pH, and injection time appeared to be the most influential for
reducing turbidity (Fig. 2a). COD and color removals were affected by
Table 5
pressure, flow rate, coagulant dosage, pH, and injection time (Fig. 2b
ANOVA of quadratic model for turbidity (Y1), COD (Y2), color (Y3) and NH-N3 (Y4) and c). However, only injection time and coagulation dose were
percentage removal with the operating parameters. (pressure (A), flow rate (B), dosage important for NH3-N removal (Fig. 2d).
of coagulant (C), pH (D) and injection time (E)). Coagulant dosage is the only factor that exhibited a significant
Source/operating Sum of Degree of Mean F Value Prob N F effect for all parameters. For example, turbidity is removed with
parameters squares freedom square increasing FeCl3 concentration up to a threshold level (above 813 g/L
Model (Y1) 49461.46 4 12365.37 20.47 b0.0001 based on Table 4), after which higher dosages produced bigger and
C 9893.44 1 9893.44 16.38 0.0002 heavier flocs. This result was likely due to high concentrations of
D 6174.92 1 6174.92 10.22 0.0025 humic acids in the leachate [34], which are able to react with the
E 7594.30 1 7594.30 12.57 0.0009 metal coagulant and form complex substances that produce sludge
2
D 25798.81 1 25798.81 42.70 b0.0001
Residual 27187.12 45 604.16
[35]. This sludge cannot be removed by DAF, thus increasing turbidity
Lack of Fit 26714.91 38 703.02 10.42 0.0017 and impeding leachate treatment [14]. COD and color exhibited a
Pure error 472.20 7 67.46 similar response to coagulant dosage (Fig. 2b and c). These results
Std. Dev. = 24.58 PRESS = 34,356.53 C.V. = 898.97 R-squared = 0.6453 suggest that similar components of the leachate contribute to both
Adj R-squared = 0.6138 Adeq precision = 14.311
COD and color [3,36]. Color was steadily removed with increasing
Model (Y2) 10469.67 8 1308.71 156.92 b0.0001 FeCl3 dose until it reached a maximum removal percentage (~96%;
A 159.37 1 159.37 19.11 b0.0001 Fig. 2c). The color of the treated leachate then increased with higher
C 2986.03 1 2986.03 358.03 b0.0001 coagulant dosage due to the excess ferric chloride in the treated
D 297.54 1 297.54 35.68 b0.0001 sample. NH3-N was also removed with increasing coagulant dosage
E 3077.06 1 3077.06 368.94 b0.0001
C2 361.33 1 361.33 43.32 b0.0001
(Fig. 2d). According to Dempsey [37], charged NH3-N particles are
D2 328.42 1 328.42 39.38 b0.0001 neutralized during coagulation and subsequently adsorbed onto floc
BC 37.69 1 37.69 4.52 0.0396 surfaces. The sludge that is generated is then removed by DAF.
CE 161.69 1 161.69 19.39 b0.0001 As shown in Fig. 2(a), (b) and (c), pH strongly influenced turbidity,
Residual 341.95 41 8.34
COD and color removals. Coagulation induced by FeCl3 was most
Lack of Fit 331.49 34 9.75 6.52 0.0076
Pure Error 10.46 7 1.49 effective under acidic conditions. This finding is consistent with
Std. Dev. = 2.89 PRESS = 543.48 C.V. = 4.99 R-squared = 0.9684 previous leachate treatment studies that used coagulation/flocculation
Adj R-squared = 0.9622 Adeq precision = 45.980 strategies [6,10,33]. This is likely due to the adsorption of negatively
charged organic matter [21] onto the positively charged coagulant ions
Model (Y3) 14882.44 7 2126.06 50.37 b0.0001
A 249.43 1 249.43 5.91 0.0194
(e.g., Fe3+, Fe(OH)+2 , and Fe(OH)
2+
) under acid conditions [38]. The
C 5577.22 1 5577.2 132.14 b0.0001 ideal pH for leachate treatment thus lies in the slightly acidic region
D 177.80 1 177.80 4.21 0.0464 (pH = 4.76). On the other hand, pH did not appear to influence NH3-N
E 1856.22 1 1856.22 43.98 b0.0001 removal (Fig. 2d). At pH 4 to 7, the majority of ions will exist as NH+4
2
C 689.87 1 689.87 16.34 0.0002
[39], thus the pH range (4 to 6) considered for this study did not alter
D2 464.33 1 464.33 11.00 0.0019
CD 797.30 1 797.30 18.89 b0.0001 conditions enough to affect NH3-N adsorption.
Residual 1772.73 42 42.21 Injection time also influenced turbidity, COD, color and NH3-N
Lack of Fit 1684.84 35 48.14 3.83 0.0350 reduction, as all variables were significantly reduced with increasing
Pure Error 87.89 7 12.56 injection time (Fig. 2(a), (b), (c) and (d)). This may be due to the
Std. Dev. = 6.50 PRESS = 2627.93 C.V. = 8.80 R-squared = 0.8936
Adj R-squared = 0.8758 Adeq precision = 24.112
dilution effect resulting from the sustained injection of the air-
saturated water.
Model (Y4) 8429.14 3 2809.71 94.90 b0.0001 Overall, pressure and flow rate did not appear to be as important as
C 620.60 1 620.60 20.96 b0.0001 other factors. Fig. 2b and c indicate that slight increases in these
E 7473.73 1 7473.73 252.44 b0.0001
parameters resulted in slightly enhanced removal of COD and color.
E2 334.81 1 334.81 11.31 0.0016
Residual 1361.89 46 29.61 However, saturator pressure and flow rate are very important in order
Lack of Fit 1185.24 39 30.39 1.20 0.4314 to obtain higher saturator efficiency and bubble volume concentration
Pure error 176.66 7 25.24 as well as smaller bubble size [16]. These features are important for
Std. Dev. = 5.44 PRESS = 1617.29 C.V. = 17.13 R-squared = 0.8609 designing and operating of DAF as a solid–liquid separation process
Adj R-squared = 0.8518 Adeq precision = 24.819
[18].
 M.N. Adlan et al. / Desalination 277 (2011) 74–82 79

Fig. 2. Perturbation plot for (a) turbidity, (b) color, (c) COD, and (d) NH3-N removal. Coded values are shown for each factor and refer to actual values listed in Table 3 (Note: A =
pressure, B = flow rate, C = dosage, D = pH and E = injection time).

In order to study the interactive relationship between independent shows that at constant pH 5, the turbidity removal has reduced with an
variable and responses, 3D surface response and counter plots of the increase in coagulant dosage.
quadratic model were drawn using Design Expert software. Fig. 3(a), Based on Fig. 3(b), the COD removal was optimum at higher dosage
(b), (c) and (d), show the 3D surface response and counter plots, where of 812.5 to 1156.25 mg/L FeCl3, with pH between 4.5 and 5.5. At this
two variables were varied within the experimental range while others condition, the percentage removal of COD was 73%. In this figure, the
were kept constant. The constant variables were chosen based on the dosage and the pH values were varied, while other factors (pressure,
level of sensitivity of the variables toward the responses, (based on flow rate and injection time) were kept constant at higher values due to
Fig. 2). the same reason as for the turbidity removal as shown in Fig. 3(b).
Fig. 3(a), shows the pH and dosages of coagulant were varied. Other However, for COD removal pressure and flow rate were observed given
factors pressure, flow rate and injection time were kept constant at the less significant effect compared to Fig. 2(a).
maximum values. This is because, based on the perturbation plots, only As for color removal, Fig. 3(c) indicated that the maximum removal
three factors (pH, coagulant dosage and injection time) had contributed was at coagulant dosage of 812.5 mg/L with pH value between 4.5 and 5.
significant effect toward the percentage removal of turbidity. As Again in this 3D response surface, the coagulant dosage and pH were
mentioned above, highest injection time, flow rate and pressure offer varied. This is because, the perturbation plot in Fig. 2(c), suggested that
a favorable condition. Due to that, these three conditions were kept coagulant dosage and pH had significant effect toward the color removal.
constant at highest value (120 s, 6 L/min and 600 kPa respectively). In terms of ammonia nitrogen removal Fig. 3(d) indicates that, the
Based on Fig. 3(a), the optimum condition was at pH 5, with coagulant dosage and the injection time were varied based on the significance in
dosage 125 mg/L, and the removal of turbidity was 70%. Fig. 3(a) also the perturbation plot (Fig. 2(d)). Other factors (pressure, flow rate
80 M.N. Adlan et al. / Desalination 277 (2011) 74–82

Fig. 3. 3D response surface for (a) turbidity, (b) COD, (c) color and (d) ammonia nitrogen removal.

Fig. 4. Overlay defining optimal treatment conditions.

 M.N. Adlan et al. / Desalination 277 (2011) 74–82 81

and pH) were kept constant. The highest removal was obtained at the Acknowledgments
highest dosage with highest injection time. At this condition, the
ammonia nitrogen removal was 49%. This study was funded by the Ministry of Science, Technology
and Environment of Malaysia (Grant no. 6013309) and Universiti
Sains Malaysia USM-RU-PGRS (Grant no. 1001/PAWAM/8042021).
3.3. Optimization of experimental conditions The author also wishes to acknowledge cooperation provided by the
Majlis Perbandaran Seberang Perai, Penang and the contractor Idaman
Graphical optimization was used to determine the optimum process Bersih Sdn. Bhd., Penang
parameters for maximum removal of turbidity, COD, color, and NH3-N
from leachate using the combined coagulant-DAF process (Fig. 4). The References
contour line for response surfaces were superimposed in an overlay plot.
By defining the value of the responses, the shaded portion of the overlay [1] N. Aghamohammadi, H.A. Aziz, M.H. Isa, A.A. Zinatizadeh, Powdered activated
carbon augmented activated sludge process for treatment of semi-aerobic landfill
plot as shown in Fig. 4, defines the permissible values of the dependent
leachate using response surface methodology, Bioresour. Technol. 98 (2007)
values. The optimum region was identified by considering the highest 3570–3578.
turbidity, COD, color, and NH3-N removals. Three independent variables, [2] I. Holmes, Dumping, burning and landfill, Issues Environ. Sci. Technol. 27 (2009)
pressure, flow rate and injection time were kept constant while dosage 75–90.
[3] Z.P. Wang, Z. Zhang, Y.J. Lin, N.S. Deng, T. Tao, K. Zhuo, Landfill leachate treatment
and pH, were used in x and y-axes to construct the graph. These two by a coagulation-photooxidation process, J. Hazard. Mater. 95 (2002) 153–159.
independent variables were varied because; coagulant dosage and pH [4] A.I. Zouboulis, X.L. Chai, I.A. Katsoyiannis, The application of bioflocculant for the
were highly influenced the dependent variables. removal of humic acids from stabilized landfill leachates, J. Environ. Manage. 70 (2004)
35–41.
Using this overlay plot few points were selected in the optimum [5] I. Ozturk, M. Altinbas, I. Koyuncu, O. Arikan, C. Gomec-Yangin, Advanced physico-
region to obtain optimum value for the responses as well as value for the chemical treatment experiences on young municipal landfill leachates, Waste
varied independent variables (coagulant dosage and pH). Based on Manage. 23 (2003) 441–446.
[6] X. Ntampau, A.I. Zoubolis, P. Samaras, Appropriate combination of physico-
Fig. 4, the optimum region as shown by point A, occurred at pressure chemical methods (coagulation/flocculation and ozonation) for the efficient
(600 kPa), FeCl3 dosage (547.66 mg/L), pH (4.88), flow rate (6 L/min) treatment of landfill leachates, Chemosphere 62 (2006) 722–730.
and injection time (98 s). The removal percentages of turbidity, COD, [7] H.A. Aziz, S. Alias, M.N. nAdlan, F.A.H. Asaari, M.S. Zahari, Colour removal from landfill
leachate by coagulation and flocculation processes, Bioresour. Technol. 98 (2007)
color and NH3-N predicted by the model using Eq. (1)–Eq. (4) (i.e., 51%,
218–220.
73%, 91%, and 40%, respectively) were in good agreement with empirical [8] T.G. Bulc, Long term performance of a constructed wetland for landfill leachate
results using these optimum conditions (Table 6). treatment, Ecol. Eng. 26 (2006) 365-374.
[9] P. Palaniandy, M.N. Adlan, H.A. Aziz, M.F. Murshed, Application of dissolved air flotation
However, the values of the factors particularly coagulant dosage
(DAF) in semi-aerobic leachate treatment, Chem. Eng. J. 157 (2010) 316–322.
and injection time could influence the studied responses (run 3, [10] A. Amokrane, C. Comel, J. Veron, Landfill leachates pretreatment by coagulation-
Table 4). This was due to higher coagulant dosage and injection time flocculation, Wat. Res. 31 (1997) 2775–2782.
where the latter may contribute to the dilution effect of the response [11] C. Di Iaconi, R. Ramadori, A. Lopez, Combined biological and chemical degradation
for treating a mature municipal landfill leachate, Biochem. Eng. J. 31 (2006)
variables. Thus the optimization process in this study successfully 118–124.
reduced the coagulant dosage and injection time which consequently [12] X. Wang, S. Chen, X. Gu, K. Wang, Pilot study on the advanced treatment of landfill
result in lower sludge production and less dilution effect [9]. leachate using a combined coagulation, fenton oxidation and biological aerated
filter process, Waste Manage. 29 (2009) 1354–1358.
[13] J. Rubio, M.L. Souza, R.W. Smith, Overview of flotation as a wastewater treatment
technique, Miner. Eng. 15 (2002) 139–155.
4. Conclusions [14] A.A. Al-Shamrani, A. James, H. Xiao, Separation of oil from water by dissolved air
flotation, Colloids Surf. A. 209 (2002) 15–26.
[15] A.A. Al-Shamrani, A. James, H. Xiao, Destabilisation of oil-water emulsion and
Leachate treatment using batch coagulation/DAF techniques was separation by dissolved air flotation, Water Res. 36 (2002) 1503–1512.
optimized using RSM. The influence of five decisive factors (i.e., pressure, [16] C.C. Ross, B.M. Smith, G.E. Valentine, Rethinking dissolved air flotation (DAF)
design for industrial pre-treatment, WEF and Purdue University Industrial Wastes
flow rate, coagulant dose, pH and injection time) were modeled and
Technical Conference, 2000.
optimized to increase the removal of four response variables. The [17] S.E. Burns, S. Yiacoumi, C. Tsouris, Microbubble generation for environmental and
optimum conditions suggested from the model for pressure, flow rate, industrial separation, Sep. Purif. Technol. 11 (1997) 221–232.
[18] L.A. Feris, J. Rubio, Dissolved air flotation (DAF) performance at low saturation
coagulant dosage, pH and injection time were 600 kPa, FeCl3 dosage of
pressure, Filtration and Separation Conference, 1999.
547.66 mg/L, pH 4.88, flow rate 6 L/min and injection time 98 s [19] A.I. Zouboulis, A. Avranas, Treatment of oil-in-water emulsions by coagulation and
respectively. Under these conditions, results from laboratory tests showed dissolved-air-flotation, Colloids Surf. A. 172 (2000) 152–161.
that removals of turbidity, COD, color and NH3-N were 53%, 72%, 94% and [20] HDR Engineering Inc., Handbook of Public Water System, John Wiley and Sons,
New York, 2001.
43% respectively. Overall, pressure and flow rate were less significant for [21] A.I. Zoubolis, W. Jun, I.A. Katsoyiannis, Removal of humic acids by flotation,
pollutant removal; however, these two factors are fundamental for Colloids Surf. A. 231 (2003) 181–193.
maximizing the performance of the DAF system. These findings are [22] S.P. Chai, R. Zakaria, Investigation on combustion characteristic of municipal solid
waste from Penang State Malaysia, International Conference on Combustion,
directly applicable to the commercial use of DAF for the treatment of Incineration/Pyrolysis and Emission Control: Waste to Wealth, 2006, Kyoto, Japan.
landfill leachate. [23] S. Ghafari, H.A. Aziz, M.H. Isa, A.A. Zinatizadeh, Application of response surface
methodology (RSM) to optimize coagulation–flocculation treatment of leachate
using poly-aluminium chloride (PAC) and alum, J. Hazard. Mater. 163 (2009)
650–656.
[24] APHA, AWWA, WPCF, Standard Methods for the Examination of Water and
Table 6 Wastewater, 21st ed American Public Health Association, Washington, D.C., USA,
Removal of leachate pollutants using model-predicted optimum values of pressure 2005.
(600 kPa), FeCl3 dosage (547.66 mg/L), pH (4.88), flow rate (6 L/min) and injection [25] M.N.B. Adlan, M.F. Murshed, H.A. Aziz, Pilot plant study on DAF process for
time (98 s). removal of turbidity and suspended solids, Proceeding Asia Water Conference,
2006, K. Lumpur.
Responses [26] D.C. Montgomery, Design and Analysis of Experiments, seventh ed. John Wiley &
Sons, New York, 2009.
Turbidity COD Color NH3-N
[27] P. Palaniandy, M.N. Adlan, H.A. Aziz, Dissolved air flotation (DAF) process for colour
Experimental value 53% 72% 94% 43% and chemical oxygen demand (COD) removal in landfill leachate treatment,
Model response 51% 73% 91% 40% Proceeding of International Conference on Chemical, Biological, and Environmental
Error 2 1 3 3 Engineering, 2009, Singapore.
[28] J.K. Edzwald, J.P. Walsh, Dissolved Air Flotation: Laboratory and Pilot Plant
Standard deviation 1.41 0.71 2.12 2.12
Investigation, AWWA Research Foundation, Denver, CO, 1992.
82 M.N. Adlan et al. / Desalination 277 (2011) 74–82

[29] C. Zhang, Y. Wang, Removal of dissolved organic matter and phthalic acid ester [35] N. Calace, A. Liberatori, B.M. Petronio, M. Pietroletti, Characteristic of different
from landfill leachate through a complexion–flocculation process, Waste Manage. molecular weight fractions of organic matter in landfill leachate and their role in
29 (2009) 110–116. soil sorption of heavy metals, Environ. Pollut. 113 (2001) 331–339.
[30] J. Duan, J. Gregory, Coagulation by hydrolysing metal salts, Adv. Colloid Interface [36] H.A. Moreno-Casillas, D.L. Cocke, J.A.G. Gomes, P. Morkovsky, J.R. Parga, E. Peterson,
Sci. 100–102 (2003) 475–502. Electrocoagulation mechanism for COD removal, Sep. Purif. Technol. 56 (2007)
[31] Water Environment Federation (WEF), Industrial Wastewater Management, 204–211.
Treatment, and Disposal, McGraw-Hill Professional, USA, 2008. [37] B.A. Dempsey, Coagulation characteristic and reactions, in: G. Newcombe, D.
[32] S. Verma, B. Prasad, I.M. Misha, Pretreatment of petrochemical wastewater by Dixon (Eds.), Interface Science in Drinking Water Treatment, Elsevier Ltd, 2006,
coagulation and flocculation and the sludge characteristics, J. Hazard. Mater. 178 pp. 5–24.
(2010) 1055–1064. [38] J. Duan, J. Gregory, Coagulation by hydrolyzing metals salts, Adv. Colloid Interface
[33] A.A. Tatsi, A.I. Zouboulis, K.A. Matis, P. Samaras, Coagulation–flocculation Sci. 100–102 (2003) 475–502.
pretreatment of sanitary landfill leachates, Chemosphere 53 (2003) 737–744. [39] M.I. Aguilar, J. Saez, M. Llorens, A. Soler, J.F. Ortuno, Nutrient removal and
[34] G. Annadurai, S.S. Sung, D.J. Lee, Simultaneous removal of turbidity and humic acid sludge production in the coagulation–flocculation process, Water Res. 36 (2002)
from high turbidity stormwater, Adv. Environ. Res. 8 (2004) 713–725. 2910–2919.