i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 7 8 0 1 e1 7 8 0 8

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

Effects of biomass, COD and bicarbonate concentrations on

fermentative hydrogen production from POME by granulated
sludge in a batch culture

Parviz Mohammadi a,c, Shaliza Ibrahim a, Mohamad Suffian Mohamad Annuar b,*
a
Department of Civil Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia
b
Institute of Biological Sciences, Faculty of Science, University of Malaya, 50603 Kuala Lumpur, Malaysia
c
Department of Environmental Health Engineering-Kermanshah, Health Research Center (KHRC), Kermanshah University of Medical Science,
Kermanshah, Iran

article info abstract

Article history: Effects of three selected variables viz. biomass concentration, initial chemical oxygen
Received 3 April 2012 demand (COD) concentration and initial bicarbonate alkalinity (BA) on biological hydrogen
Received in revised form production from palm oil mill effluent (POME) using the granulated sludge in batch culture
11 August 2012 were investigated. The experimental results were analyzed and modeled using a central
Accepted 24 August 2012 composite design (CCD) of response surface methodology (RSM). In order to carry out
Available online 26 September 2012 a comprehensive analysis of the biohydrogen production process, indicative parameters
namely hydrogen yield (YH), specific hydrogen production rate (SHPR), and COD removal
Keywords: efficiency were studied as the process responses. Maximum hydrogen yield (124.5 mmol
Fermentative hydrogen production H2/g CODremoved) and specific hydrogen production rate (55.42 mmol H2/g VSS.d) were
Granulated sludge achieved at CODin 3000 and 6500 mg/l, MLVSS 4000 and 2000 mg/l, and initial BA 1100 mg
Palm oil mill effluent CaCO3/l, respectively.
Process factors Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
Response surface methodology reserved.

1. Introduction a colloidal suspension contained 95e96% water, 0.6e0.7% oil

One of the most important contributors to Malaysia’s
economy is palm oil industry. This industry generates large
quantities of polluted wastewater commonly called as palm
oil mill effluent (POME). It is estimated that 5e7.5 tonnes of
water are required for each tonne of crude palm oil produc-
tion; and more than 50% of the water ends up as POME [1,2].
Based on palm oil production industry in Malaysia, 15.2
million tonnes of POME were produced in 2005 alone [3]. The
three main sources of POME are sterilization (36%), clarifica-
tion (60%), and hydrocyclone (4%) units. Raw POME as
and 4e5% total solids [2,4].
Nowadays, fossil fuels are the primary source of global
energy requirements, with their foreseeable depletion due to
limited fossil energy resource. With respect to global envi-
ronmental impacts due to the usages of this energy resource,
such as greenhouse effect, ozone layer depletion and resource
recovery, intensive search is going on worldwide for renew-
able and non-polluting energy source [5e16]. Hydrogen is an
effective fuel with high-energy yield (122 kJ/g) which is 2.4, 2.8
and 4 times higher than energy yields of methane, gasoline
and coal, respectively [17e23].

* Corresponding author. Tel.: þ60 3 79674003; fax: þ60 3 79674178.

E-mail address: suffian_annuar@um.edu.my (M.S. Mohamad Annuar).
0360-3199/$ e see front matter Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijhydene.2012.08.110
17802 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 7 8 0 1 e1 7 8 0 8
Fermentation of agro-waste biomass using microorganisms
is a promising approach for producing hydrogen which is
known as the cleanest renewable energy. Fermentative
hydrogen production is influenced by many factors such as
inoculum, substrate, alkalinity, reactor type, organic loading
rate, pH and temperature. Some aforementioned parameters
have been studied before. Chong and coworkers [3] have
investigated the effect of pH on hydrogen production from
POME. The maximum hydrogen production and maximum
volumetric hydrogen production rate were achieved at 3195 ml
H2/l POME and 1034 ml H2/l h at pH 5.5 and 37  C in temperature,
respectively. Also, the maximum process stability in an anaer-
obic sequencing batch reactor (ASBR) for biohydrogen produc-
tion from POME was studied by Prasertsan et al. [24]. At an
organic loading rate (OLR) of 60 g/l d and hydraulic retention
time (HRT) of 2 days, maximum hydrogen yield of 0.27 l H2/g
COD with a volumetric hydrogen production rate of 9.1 l H2/l d
were achieved. The most researches have been carried out
investigating the effect of operational factors on fermentative
hydrogen production. While in this research, the effect of
process factors on biological hydrogen production is studied for
estimation of optimum conditions.
Response surface methodology (RSM) is a collection of
mathematical and statistical techniques useful for analyzing
the effects of several independent variables on the response
[25e27]. RSM has an important application in the process
analysis and optimization as well as the improvement of
existing design. There have been several studies the interac-
tive effect among the variables on the optimization of
fermentative hydrogen production process to depict by using
RSM. The effects of three variables (carbon to nitrogen (C/N)
ratio, carbon to phosphate (C/N) ratio and iron concentration)
in POME on fermentative hydrogen production using RSM
have been studied in an anaerobic sequencing batch reactor
operating [28]. The RSM results showed that optimal hydrogen
production and chemical oxygen demand (COD) removal were
at C/N ratio, C/P ratio, and Fe2þ of 74, 559, and 257 mg,
respectively. Also, the RSM results indicated that optimum
conditions for hydrogen production (306 ml H2/g carbohy-
drate) were pH 5.69, 36  C, and 92 g COD/l [29].
The efficiency of biohydrogen production is directly related
to the optimal control of food to microorganism (F/M) ratio
[30,31]. The metabolic and kinetic characteristics of microor-
ganisms were affected by initial COD and biomass concentra-
tion ratio significantly [31,32]. There exists certain
Table 1 e Characteristics of raw and per-settled POME.
Parameters Raw POME
Amount Standard deviation
BOD5 22,500 2150
TCOD 49,800 4250
SCOD 21,950 2320
TSS 18,800 1240
TKN 430 32
TP 84 7.5
Oil and grease 4250 62
pH 4.2 0.2
All in mg/l except pH.
disagreement on the optimal F/M ratio for fermentative
hydrogen production. It has been demonstrated that in an
appropriate range of F/M could increase the ability of hydrogen
producing bacteria to produce hydrogen during fermentative
hydrogen production, but F/M ratio at much higher levels could
decrease it with increasing levels [31].
Fermentative hydrogen production processes produce
bicarbonate alkalinity (BA) which could maintain the pH at
favorable range for hydrogenesis activity. Therefore, in this
study, the effects of three important process variables viz.
initial COD concentration, biomass concentration and initial
BA, on biological hydrogen production from POME in batch
culture were investigated. The experiments were designed as
central composite face-centered design (CCFD) using Design
Expert software (Stat-Ease Inc., version 6.0.8).
2. Materials and methods
2.1. Wastewater preparation
Raw POME samples were taken from SIME Darby Plantation
Palm Oil Industry Sdn. Bhd., Nilai, Malaysia. The samples were
transferred to the laboratory immediately and stored in a cold
room (4  C) before use. The characteristics of the raw and pre-
settled POME are summarized in Table 1. This was used as the
liquid medium in the batch fermentation study.
2.2. Seed sludge preparation
The inoculum for the batch fermentation was the granulated
sludge obtained from a working upflow anaerobic sludge
blanket-fixed film (UASB-FF) bioreactor operated at steady-
state (HRT of 12 h, influent COD 25880 mg/l, and OLR of
51.8 g COD/l d) in our laboratory for biological hydrogen
production from POME. The granulated sludge was withdrawn
from the bottom part of the reactor.
2.3. Batch experiments
Experiments were performed in 163-ml serum bottles with
50 ml medium (experimental conditions as shown in Table 2)
and sealed with butyl rubber stoppers and aluminum seals as
shown in Fig. 1. The initial pH was adjusted to 5.5 using NaOH
(3N) and HCl (3N). The medium was distributed anaerobically
Pre-settled POME
Amount Standard deviation
19,400 235
28,800 260
17,140 165
850 21
382 3.5
84 1.5
1480 13.8
4.2 0.15
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 7 8 0 1 e1 7 8 0 8 17803

Table 2 e Central composite design and experimental data.

Run Variables Responses

CODin MLVSS Alkalinity H2 yield SHPR COD removal

(mg/l) (mg/l) (mg CaCO3/l) (mmol H2/g CODrem) (mmol H2/g VSS.d) (%)

1 10,000 2000 2000 39.09 41.92 42.9

2 6500 6000 1100 62.13 17.77 52.8
3 6500 4000 1100 72.82 33.96 57.4
4 3000 2000 2000 118.46 40.42 45.5
5 10,000 2000 200 48.76 53.39 43.8
6 6500 4000 2000 58.92 25.95 54.2
7 10,000 6000 200 27.67 9.18 39.8
8 6500 2000 1100 62.92 55.42 54.2
9 6500 4000 200 69.18 28.89 51.4
10 3000 6000 2000 116.88 14.11 48.3
11 3000 2000 200 115.70 43.56 50.2
12 6500 4000 1100 69.39 34.22 60.7
13 3000 4000 1100 124.48 25.30 54.2
14 6500 4000 1100 78.57 36.64 57.4
15 3000 6000 200 107.72 12.36 45.9
16 10,000 6000 2000 52.84 24.39 55.4
17 6500 4000 1100 66.53 33.08 61.2
18 6500 4000 1100 62.35 30.19 59.6
19 10,000 4000 1100 58.86 37.08 50.4
20 6500 4000 1100 69.12 32.85 58.5

into the serum bottles under nitrogen gas purging. The gas
phase was also purged using nitrogen gas. The serum bottles
were placed in a shaking incubator (DAIHAN LABTECH Co.,
Singapore) (120 rpm) at 38  C. At 8 h interval, the composition
of gas and liquid (H2 percentage, volume of biogas produced,
alkalinity and COD) were measured. The results after two days
of fermentation were analyzed using analysis of variance
(ANOVA) (Design Expert Software).
2.4. Experimental design and mathematical model
Results obtained from the previous experiments showed that
BA was produced during anaerobic digestion of POME [33,34].
Fig. 1 e Serum bottles used for batch experiments.
Since one of the aims in this study was to determine the
optimum initial BA which could maintain the pH at favorable
range for hydrogenesis activity, therefore the range of BA was
chosen from a minimum amount of 200e2000 mg CaCO3/l.
Nasr and coworkers [35] have found that the optimum range of
F/M ratio for fermentative hydrogen production was 1e2 g
COD/g VSS. Also the highest hydrogen yield was obtained at an
F/M ratio of 1.24 g COD/g VSS from the range of 0.35e4.38 g
COD/g VSS tested [31]. Hence, a wide range of food to micro-
organism (F/M) ratio from about 0.25 to 2.5 g COD/g VSS d was
considered and examined as a criterion for determining the
range of CODin concentration (3000e10,000 mg/l) and biomass
concentration (2000e6000 mg/l) in this study.
The statistical method of factorial design of experiment
(DoE) eliminates systematic errors with an estimate of the
experimental error and also minimizes the number of experi-
ments [25,36]. In the present study, H2 production from POME in
batch culture was evaluated and optimized employing a central
composite design (CCD) of response surface methodology
(RSM). The main and interactive effects of CODin concentration,
biomass concentration and initial BA on the process responses
were explained. Table 2 shows the experimental conditions for
fermentative hydrogen production from POME based on CCD
design. The design includes of 2k factorial points augmented by
2k (the number of variables) axial points and a center point. The
levels of each variable vary from a low to high value, either
numerically expressed in absolute values or coded as 1 (low)
and 1 (high) with intermediate value of 0 (middle).
In order to evaluate the granular sludges performance in the
hydrogen production, three responses were selected and the
experimental results after two days under various process
conditions are presented in Table 2. Three-dimensional plots
were used to visualize the effects of the studied variables. The
simultaneous interactions of the variables on the responses
were also studied from these three-dimensional plots. The
17804 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 7 8 0 1 e1 7 8 0 8

coefficients of the polynomial model were obtained using Eq. (1)
[37].
Y ¼ b0 þ bi Xi þ bj Xj þ bii X2i þ bjj X2j þ bij Xi Xj þ /
where b is the regression coefficient, i is the linear coefficient
and j is quadratic coefficient. The terms of a model may be
selected or rejected based on P value at 95% confidence level.
The results were analyzed by ANOVA (Design Expert soft-
ware). The effects of simultaneous interaction of the variables
on the responses are investigated with respect to the three
dimensional plots and their respective contour plots. The
ANOVA values for response surface models are shown in
Table 3. Detail explanations for each response are presented
in the next section.
determined using a gas chromatograph (Perkin Elmer, Auto
system GC), equipped with thermal conductivity detector
(TCD) and data acquisition system with Total Chrom soft-
(1) ware. H2 content was also analyzed by GC-TCD fitted with
a 1.5 m stainless steel column (SS350A) packed with
a molecular sieve (80/100 mesh). The temperatures of the
injection port, oven and detector were 80, 200, and 200  C,
respectively. Argon was used as a carrier gas at a flow rate of
30 ml/min.
3. Results and discussion
3.1. Hydrogen yield

2.5. Analytical techniques
The parameters viz. biochemical oxygen demand (BOD), COD,
total suspended solids (TSS), volatile suspended solids (VSS),
alkalinity, total Kjeldahl nitrogen (TKN), oil and grease, and
pH were analyzed using procedures outlined in the APHA
standard methods [38]. The biogas composition was
Results of ANOVA in Table 3 showed that the regression
model for hydrogen yield was significant. The model was
reduced to squared and quadratic regressors after the elimi-
nation of non-significant model terms. The R-sqaured (R2)
value of the model was 0.9624 and was higher than that of the
model for COD removal. The main and second-order effects of
the variables (A, A2 and B2) and two-level interactions of mixed

Table 3 e ANOVA for response surface models applied.

Response Model type ANOVA

Source Sum of square DF Mean square F value Prob > F

Hydrogen Yield, (mmol H2/g CODrem) Reduced quadratic Model 14171.68 4 3542.92 97.18 <0.0001
A 12677.87 1 12677.87 347.75 <0.0001
A2 1259.60 1 1259.60 34.55 <0.0001
B2 307.64 1 307.64 8.44 0.0109
BC 212.18 1 212.18 5.82 0.0291
Residual 546.85 15 36.46
Std. Dev. ¼ 6.04 C.V. ¼ 8.15 PRESS ¼ 996.61
R2 ¼ 0.9628 Adj.R2 ¼ 0.9529 Adeq.
Precision ¼ 28.542

SHPR, (mmol H2 /g VSS.d) Reduced quadratic Model 2798.18 5 559.64 53.57 < 0.0001
A 91.32 1 91.32 8.74 0.0104
B 2462.07 1 2462.07 235.68 < 0.0001
B2 30.53 1 30.53 2.92 0.1094
C2 118.54 1 118.54 11.35 0.0046
BC 124.66 1 124.66 11.93 0.0039
Residual 146.25 14 10.45
Std. Dev. ¼ 3.23 C.V. ¼ 10.25 PRESS ¼ 368.97
R2 ¼ 0.9503 Adj.R2 ¼ 0.9326 Adeq.
Precision ¼ 25.601

COD removal, (%) Reduced quadratic Model 693.56 8 86.70 19.35 < 0.0001
A 13.92 1 13.92 3.11 0.1057
C 23.10 1 23.10 5.16 0.0443
A2 58.19 1 58.19 12.98 0.0041
B2 31.79 1 31.79 7.09 0.0221
C2 46.23 1 46.23 10.32 0.0083
AB 12.50 1 12.50 2.79 0.1231
AC 36.12 1 36.12 8.06 0.0161
BC 69.62 1 69.62 15.54 0.0023
Residual 49.30 11 4.48
Std. Dev. ¼ 2.12 C.V. ¼ 4.06 PRESS ¼ 262.24
R2 ¼ 0.9336 Adj.R2 ¼ 0.8854 Adeq.
Precision ¼ 13.116

A: CODin; B: MLVSS; C: initial bicarbonate alkalinity; DF: degree of freedom; Std. Dev: standard deviation; C.V.: coefficient of variation; Adj. R2:
adjusted R-squared; PRESS: prediction error sum of squares; Adeq. Precision: adequate precision.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 7 8 0 1 e1 7 8 0 8 17805

liquid volatile suspended solid (MLVSS) and initial bicarbonate amount of hydrogen yield was achieved at 124.5 mmol H2/g
alkalinity (BC ) were significant model terms. The regression COD removed for COD concentration of 3000 mg/l.
equation of the reduced quadratic model for the response is as The hydrogen yield obtained from this study was less than
follows: that generally obtained (1.45 mol H2/g COD) from hexose
substrate [39], but it was higher than those results were ob-
Hydrogen yield ¼ 34:55  17:8A þ 9:8A2  4:78 B2 þ 2:58 BC (2) tained from POME (19 mmol H2/g COD removed) by Krishnan
Fig. 2 showed the three-dimensional contour plots of the et al. [40] and (40 mmol H2/g COD removed) by Prasertsan et al.
model for variation in hydrogen yield, as a function of MLVSS [24], respectively.
(B) and initial BA (C) at three different substrate concentra-
tions (3000, 6500 and 10,000 mg COD/l). It can be seen from the 3.2. Specific hydrogen production rate
response surface plots (Fig. 2) that the hydrogen yield
increased upon increasing of the MLVSS at any initial BA until Table 3 showed the ANOVA results obtained for SHPR. The
the MLVSS concentrations reached to a level of 3450 mg/l (for variation of the SHPR as a function of the studied variables
initial BA 200 mg CaCO3/l), 4000 mg/l (for initial BA 1100 mg (CODin, MLVSS and initial BA) was described with a reduced
CaCO3/l), and 4550 mg/l (for initial BA 2000 mg CaCO3/l). quadratic model. The R2 value of the model for SHPR was 0.95.
Further increase in MLVSS concentration beyond these levels The main and second-order effects of the variables (A, B, B2
resulted in the decrease in the hydrogen yield for all the and C2) and two-level interactions of MLVSS and initial BA
substrate concentration tested. The interaction showed that (BC ) were significant model terms. Other model terms, which
the MLVSS and initial BA could affect significantly the were insignificant, were eliminated to simplify the model. The
hydrogen yield. According to Fig. 2, the maximum hydrogen regression equation of the reduced quadratic model for the
yields were achieved at biomass concentrations of 3460 mg/l response is as follows:
and 4510 mg/l and initial BA of 205 mg CaCO3/l and 1995 mg
CaCO3/l, respectively. Subsequently, the hydrogen yield SHPR; mmol H2 =g VSS:d ¼16:52 þ 1:51A  8:74B þ 1:54 B2
(3)
decreased as COD concentration increased. The maximum  3:04 C2 þ 1:97BC

a b
125.2 69.7
H2 Yield (mmol H2/g COD)

H2 Yield (mmol H2/g COD)

121.3 65.8
117.4 61.9
113.4 58.0
109.5 54.1

2000 2000
6000 6000
1550 1550
5000 5000
1100 1100
4000 4000
C: BA (mg/l) 650 C: BA (mg/l) 650
3000 B: MLVSS (mg/l) 3000 B: MLVSS (mg/l)
200 2000 200 2000

c
54.0
H2 Yield (mmol H2/g COD)

50.0
46.1
42.2
38.3

2000
6000
1550
5000
1100
4000
C: BA (mg/l) 650 3000 B: MLVSS (mg/l)
200 2000

Fig. 2 e Three-dimensional contour plots of the model for hydrogen yield (mmol H2/g COD removed) as a function of initial
BA and MLVSS at COD concentration of 3000 mg/l (a), 6500 mg/l (b) and 10,000 mg/l (c).
17806 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 7 8 0 1 e1 7 8 0 8

Fig. 3 showed the three-dimensional contour plots of the the studied variables (CODin, MLVSS and initial BA) was
model for variation in SHPR as a function of MLVSS (B) and described using a reduced quadratic model (P < 0.0001). The
initial BA (C) at three different COD concentrations within the main effects, second-order effects and the two-level interac-
design space. It is shown that by decreasing the MLVSS at the tions of CODin, MLVSS, and initial BA (A, C, A2, B2, C2, AB, AC,
middle of initial BA, more favorable environment for and BC ) were significant model terms. Only factor B was not
fermentative hydrogen production could be provided. Low a significant model term and it was eliminated to simplify the
SHPR at high MLVSS concentrations were mainly due to high model.
concentration of VSS from one side and low amount of CODin The coefficient of correlation i.e. R2 ¼ 0.9336 showed that
from another side as SHPR is expressed as mmol H2/g VSS.d. It only an insignificant 6.64% of the total variation was not
can be observed in Fig. 3aec, at the same MLVSS and initial BA, explained by the model (as R2 is a ratio of sum of square
greater SHPR was obtained from the condition with high regression to total sum of square). The high value for adjusted
substrate concentration (as shown in Fig. 3c). R2 (0.8854) also supported the significance of the model. The
Maximum SHPR was predicted at 55.43 mmol H2/g VSS d reasonable precision and reliability of the experiments was
under MLVSS, initial BA and CODin concentration of 2000 mg demonstrated by a low variation coefficient value (4.06)
VSS/l, 830 mg CaCO3/l and 10,000 mg COD/l, respectively. In (Kuehl, 2000). The regression equation of the reduced
contrast, minimum SHPR (8.05 mmol H2/g VSS d) was obtained quadratic model for the response is as follows:
under the lowest initial BA (200 mg CaCO3/l) and the highest
MLVSS (6000 mg/l) with CODin of 3000 mg/l. It is concluded COD removal; % ¼58:24  1:18A þ 1:52C  4:6A2  3:4 B2
that high specific hydrogenesis activity could be achieved at (4)
 4:1C2 þ 1:25AB þ 2:12AB þ 2:95 BC
intermediate initial bicarbonate alkalinity value and high F/M
concentration ratio. Fig. 4 showed the three-dimensional contour plots of the
model for variation in COD removal efficiency as a function of
3.3. COD removal efficiency CODin (A) and MLVSS (B) at three different values of initial BA
i.e. 200 mg CaCO3/l (a), 1100 mg CaCO3/l (b) and 2000 mg
Table 3 showed the ANOVA results obtained for COD removal CaCO3/l (c). The maximum COD removal efficiency (>58.3%)
efficiency. The variation of the COD removal as a function of was obtained at CODin, MLVSS and initial BA of 6024 mg/l,

a b
49.4 52.4
SHPR (mmol H2/g VSS.d)

SHPR (mmol H2/g VSS.d)

38.9 41.9
28.4 31.4
17.9 20.9
7.4 10.4

2000 2000
6000 6000
1550 1550
5000 5000
1100 1100
4000 4000
C: BA (mg/l) 650 C: BA (mg/l) 650
3000 B: MLVSS (mg/l) 3000 B: MLVSS (mg/l)
200 2000 200 2000

c
55.5
SHPR (mmol H2/g VSS.d)

45.0
34.4
23.9
13.4

2000
6000
1550
5000
1100
4000
C: BA (mg/l) 650 3000 B: MLVSS (mg/l)
200 2000

Fig. 3 e Three-dimensional contour plots of the model for SHPR (mmol H2/g VSS d) as a function of initial BA and MLVSS at
COD concentration of 3000 mg/l (a), 6500 mg/l (b) and 10,000 mg/l (c).
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 7 8 0 1 e1 7 8 0 8 17807

a b
54.1 58.3
COD Removal (%) 50.5 55.7

COD Removal (%)

46.8 53.1
43.2 50.4
39.6 47.8

6000 6000
10000 10000
5000 5000
8250 8250
4000 4000
6500 6500
B: MLVSS (mg/l) 3000 B: MLVSS (mg/l) 3000
4750 A: CODin (mg/l) 4750 A: CODin (mg/l)
2000 3000 2000 3000

c
56.4
53.4
COD Removal (%)

50.4
47.4
44.4

6000
10000
5000
8250
4000
6500
B: MLVSS (mg/l) 3000 4750 A: CODin (mg/l)
2000 3000

Fig. 4 e Three-dimensional contour plots of the model for COD removal as a function of CODin and MLVSS at initial BA of
200 mg CaCO3/l (a), 1100 mg CaCO3/l (b), 2000 mg CaCO3/l (c).

3956 mg/l and 1100 mg CaCO3/l, respectively. The minimum 1100 mg CaCO3/L, respectively. The maximum specific
COD removal efficiency (39.6%) was obtained at CODin of hydrogen production rate (55.42 mmol H2/g VSS.d) was
10,000 mg/l, MLVSS of 6000 mg/l and initial BA of 200 mg observed at the CODin, MLVSS and initial BA of 6500 mg/l,
CaCO3/l. 2000 mg/l and 1100 mg CaCO3/L, respectively.
The drop in pH below 5.0 observed earlier in the UASB-FF
(Mohammadi et al., submitted) was attributed to the high
production of volatile fatty acids (VFAs) at the beginning of the references
process and insufficient bicarbonate alkalinity to buffer the
acid production. The results obtained from this part of the
study can be used to project the required amount of initial BA
[1] Ahmad AL, Ismail S, Bhatia S. Water recycling from palm oil
for effective buffering action in the UASB-FF based on the mill effluent (POME) using membrane technology.
influent CODin as well as the recycle ratio. Desalination 2003;157:87e95.
[2] Zinatizadeh AAL, Mohamed AR, Abdullah AZ, Mashitah MD,
Hasnain Isa M, Najafpour GD. Process modeling and analysis
4. Conclusions of palm oil mill effluent treatment in an up-flow anaerobic
sludge fixed film bioreactor using response surface
methodology (RSM). Water Res 2006;40:3193e208.
Effects of three important variables viz. initial COD concen- [3] Chong ML, Abdul Rahim R, Shirai Y, Hassan MA. Biohydrogen
tration (CODin), biomass concentration and initial BA on the production by Clostridium butyricum EB6 from palm oil mill
biological activity of the granulated sludge in Hydrogen effluent. Int J Hydrogen Energy 2009;34:764e71.
production in batch culture were investigated and modeled [4] Ma AN. Environment management for the palm oil industry.
using a central composite design (CCD) of response surface Palm Oil Develop 2000;30:1e10.
[5] Lin CY, Lay CH. Carbon/nitrogen-ratio effect on fermentative
methodology (RSM). Hydrogen yield (YH), specific hydrogen
hydrogen production by mixed microflora. Int J Hydrogen
production rate (SHPR), and COD removal efficiency were
Energy 2004;29:41e5.
studied as the simultaneous responses. The maximum [6] Mohammadi P, Ibrahim S, Mohamad Suffian MA, Sean L.
hydrogen yield (124.5 mmol H2/g CODremoved) was obtained at Effects of different pretreatment methods on anaerobic
the CODin, MLVSS and initial BA of 3000 mg/l, 4000 mg/l and mixed microflora for hydrogen production and COD
17808 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 7 8 0 1 e1 7 8 0 8
reduction from palm oil mill effluent. J Clean Production
2011;19:1654e8.
[7] Hart. Hydrogen power. The commercial future of the “ultimate
fuel”. London: Financial Times Energy Publishing; 1997.
[8] Najafpour GD, Younesi H. Bioconversion of synthesis gas to
hydrogen using a light-dependent photosynthetic bacterium,
Rhodospirillum rubrum. World J Microbiol Biotechnol 2007;23:
275e84.
[9] Mohan SV, Mohanakrishna G, Purushotham Reddy B,
Saravanan R, Sarma PN. Bioelectricity generation from
chemical wastewater treatment in mediatorless (anode)
microbial fuel cell (MFC) using selectively enriched hydrogen
producing mixed culture under acidophilic
microenvironment. Biochem Eng J 2008;39:121e30.
[10] Najafpour G, Younesi H, Mohamed AR. Effect of organic
substrate on hydrogen production from synthesis gas using
Rhodospirillum rubrum, in batch culture. Biochem Eng J 2004;
21:123e30.
[11] Younesi H, Najafpour G, Ismail KSK, Mohamed AR,
Kamaruddin AH. Biohydrogen production in a continuous
stirred tank bioreactor from synthesis gas by anaerobic
photosynthetic bacterium: Rhodopirillum rubrum. Bioresour
Technol 2008;99:2612e9.
[12] Yu H, Zhu Z, Hu W, Zhang H. Hydrogen production from rice
winery wastewater in an upflow anaerobic reactor by using
mixed anaerobic cultures. Int J Hydrogen Energy 2002;27:
1359e65.
[13] Fan YT, Li CL, Lay JJ, Hou HH, Zhang GS. Optimization of
initial substrate and pH levels for germination of sporing
hydrogen producing anaerobes in cow dung compost.
Bioresour Technol 2004;91:189e93.
[14] Najafpour G, Younesi H, Mohamed A. A survey on various
carbon sources for biological hydrogen production via the
water-gas reaction using a photosynthetic bacterium
(Rhodospirillum rubrum). Energy Sources Part A Recovery
Utiliz Environ Effects 2006;28:1013e26.
[15] Kapdan IK, Kargi F. Bio-hydrogen production from waste
materials. Enzyme Microb Technol 2006;38:569e82.
[16] Ginkel SWV, Oh SE, Logan BE. Biohydrogen gas production
from food processing and domestic wastewaters. Int J
Hydrogen Energy 2005;30:1535e42.
[17] Atif AAY, Razia AF, Ngan MA, Morimoto M, Iyukec SE,
Veziroglu NT. Fed batch production of H2 from palm oil mill
effluent using anaerobic microflora. Water Sci Technol 2005;
53:271e9.
[18] Mohan SV, Lalit VB, Sarma PN. Anaerobic biohydrogen
production from dairy wastewater treatment in sequencing
batch reactor (AnSBR): effect of organic loading rate. Enzyme
Microb Technol 2007;41:506e15.
[19] Wu KJ, Chang JS. Batch and continuous fermentative production
of hydrogen with anaerobic sludge entrapped in a composite
polymeric matrix. Process Biochem 2007;42:279e84.
[20] Najafpour G, Ismail KSK, Younesi H, Mohamed AR,
Kamaruddin AH. Performance of biological hydrogen
production process from synthesis gas, mass transfer in
batch and continuous bioreactors. Int J Eng Trans B Appl
2004;17:105e20.
[21] Zhang ZP, Show KY, Tay JH, Liang DT, Lee DJ. Biohydrogen
production with anaerobic fluidized bed reactors e
a comparison of biofilm-based and granule-based systems.
Int J Hydrogen Energy 2008;33:1559e64.
[22] Leite JAC, Fernandes BS, Pozzi E, Barboza M, Zaiat M.
Application of an anaerobic packed-bed bioreactor for the
production of hydrogen and organic acids. Int J Hydrogen
Energy 2008;33:579e86.
[23] Zadariana J, Mohamad SMA, Ibrahim S, Vikineswary S.
Optimization of phototrophic H2 production by
Rhodopseudomonas palustris PBUM001 via statistical
experimental design. Int J Hydrogen Energy 2009;34:7502e12.
[24] Prasertsan P, Sompong OT, Birkeland NK. Optimization and
microbial community analysis for production of biohydrogen
from palm oil mill effluent by thermophilic fermentative
process. Int J Hydrogen Energy 2009;34:7448e59.
[25] Zinatizadeh AAL, Mohamed AR, Mashitah MD, Abdullah AZ,
Hasnain Isa M. Optimization of pre-treated palm oil mill
effluent digestion in an up-flow anaerobic sludge fixed film
bioreactor: a comparative study. Biochem Eng J 2007;35:
226e37.
[26] Akhbari A, Zinatizadeh AA, Mohammadi P, Irandoust M,
Mansouri Y. Process modeling and analysis of biological
nutrients removal in an integrated RBC and AS system using
response surface methodology. Chem Eng J 2011;168:269e79.
[27] Mansouri Y, Zinatizadeh AA, Mohammadi P, Irandoust M,
Akhbari A. Hydraulic characteristics analysis of an anaerobic
rotatory biological contactor (AnRBC) using tracer
experiments and response surface methodology (RSM).
Korean J Chem Eng 2012;29:891e902.
[28] Sompong OT, Prasertsan P, Intrasungkha N,
Dhamwichukorn S, Birkeland NK. Optimization of
simultaneous thermophilic fermentative hydrogen
production and COD reduction from palm oil mill effluent by
Thermoanaerobacterium-rich sludge. Int J Hydrogen Energy
2008;33:1221e31.
[29] Chong ML, Rahman NAA, Rahim RA, Suraini AA, Shirai Y,
Hassan MA. Optimization of biohydrogen production by
Clostridium butyricum EB6 from palm oil mill effluent using
response surface methodology. Int J Hydrogen Energy 2009;
34:7475e82.
[30] Chen WH, Chen SY, Khanal SK, Sung S. Kinetic study of
biological hydrogen production by anaerobic fermentation.
Int J Hydrogen Energy 2006;31:2170e8.
[31] Chakkrit S, Tsuyoshi I, Pensri P, Alissara R. Optimization of
key factors affecting hydrogen production from food waste
by anaerobic mixed cultures. Int J Hydrogen Energy 2011;36:
14120e33.
[32] Liu Y. Bioenergetic interpretation on the So/Xo ratio in
substrate-sufficient batch culture. Water Res 1996;30:
2766e70.
[33] Mohammadi P, Ibrahim S, Annuar MSM. Optimization of
fermentative hydrogen production from palm oil mill
effluent in an up-flow anaerobic sludge blanket fixed film
bioreactor. J Power Sources, submitted for publication.
[34] Mohammadi P, Ibrahim S, Annuar MSM, Ghafari S,
Vikineswary S, Zinatizadeh AA. Influences of environmental
and operational factors on dark fermentative hydrogen
production: a review. Clean Soil Air Water, in press.
[35] Nasr N, Elbeshbishy E, Hafez H, Nakhla G, Hesham M,
Naggar EL. Biohydrogen production from thin spillage using
conventional and acclimatized anaerobic digester sludge. Int
J Hydrogen Energy 2011;36:12761e9.
[36] Kuehl RO. Design of experiments: statistical principles of
research design and analysis. 2nd ed. Pacific Grove: Duxbury
Press; 2000. pp. 2e225.
[37] Khuri AI, Cornell JA. Response surfaces: design and analyses.
2nd ed. New York: Marcel Dekker; 1996. pp. 80e170.
[38] American Public Health Association (APHA). Standard
methods for the examination of water and wastewater. 20th
ed. Washington DC: APHA, AWWA; 1999.
[39] Zhao BH, Yue ZB, Zhao QB, Mu Y, Yu HQ, Harada H, et al.
Optimization of hydrogen production in a granule-based
UASB reactor. Int J Hydrogen Energy 2008;33:2454e61.
[40] Krishnan V, Desa A. Biohydrogen generation from palm oil
mill effluent using anaerobic contact filter. Int J Hydrogen
Energy 2006;31:1284e91.