ARTICLE IN PRESS

Fuel xxx (2010) xxxxxx

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Kinetic study of hydroxide-catalyzed methanolysis of Jatropha curcaswaste food

oil mixture for biodiesel production
Hanny Johanes Berchmans *, Kayoko Morishita, Takayuki Takarada
Department of Chemical and Environmental Engineering, Gunma University, 1-5-1 Tenjin-cho, Kiryu-shi, Gunma 3768515, Japan

a r t i c l e

i n f o

Article history:
Received 19 September 2009
Received in revised form 2 December 2009
Accepted 11 January 2010
Available online 25 January 2010
Keywords:
Kinetic
Methanolysis
Waste food oil
Jatropha curcas oil
Free fatty acids

a b s t r a c t
Hydroxide-catalyzed methanolysis kinetic reaction model that considers the side effects of saponication
of glycerides, esters and free fatty acids (FFAs) was proposed. Methanolysis of a mixture of Jatropha curcaswaste food oil that contains 1%wt FFAs under reaction temperature 50 C, mixing speed 900 rpm,
methanol to oil molar ratio 6:1 and KOH to oil mass ratio 1%wt, was accurately and completely described
by the kinetic reaction model. Simulation and experiment results showed that the FFAs were neutralized
by the hydroxide ion, resulting in production of water and more soap. This was undesired side reaction
because it consumed more catalyst to achieve the same reaction rate. However, 1%wt FFAs content in the
oil mixture and 2 h reaction time of methanolysis produced biodiesel with 97.1%wt purity.
2010 Elsevier Ltd. All rights reserved.

1. Introduction
Many research studies had been carried out to seek the alternative ways to reduce the biodiesel production cost through the
application of more-effective production technologies and the utilization of less expensive vegetable oils. Minimizing the biodiesel
production cost by using inexpensive and inedible feedstock such
as Jatropha curcas, waste animal fats, waste food oil and its mixtures is more practical because inedible vegetable oils and waste
vegetable oils are readily available and also not competing with
food oils. An implementation of mixture of J. curcas oil with inexpensive waste food oil as biodiesel feedstock is an interesting
choice for regions that have insufcient fresh vegetable oils resource but have signicant amount of waste food oil. However,
the main problem on the using those oils is free fatty acids (FFAs)
content within the oils when alkali based catalyzed alcoholysis
process was used [1,2]. High FFAs content (>1%wt) caused soap formation during alcoholysis process and lead to the difculties in
separation of biodiesel from its by product, and as a result it reduced the yield of conversion [3]. To overcome this problem, it is
possible to control the FFAs content of J. curcas oil by mixing it with
rened waste food oil that has lower FFAs. Beside the using inexpensive feedstock, researchers are focusing to minimize production cost by improving the production technologies. One of those
efforts is maximizing the yield of alkali catalyzed alcoholysis reaction by controlling the reaction parameters that affect the reaction
* Corresponding author. Tel.: +81 277 30 1452; fax: +81 277 30 1454.
E-mail address: hannyjberchmans@gmail.com (H.J. Berchmans).

rate and mechanism. A kinetic study of alcoholysis describes the

reaction mechanism and provides the parameters that can be used
to predict the compositions and the concentration of the reaction
mixture at any time under certain reaction conditions. There are
a number of kinetics studies on vegetable oils alcoholysis in literature. Freedman [4] studied the kinetics of the acid and base-catalyzed alcoholysis of soybean oil with 1-butanol and methanol at
30:1 and 6:1 M ratio of alcohol to oil. Noureddini and Zhu [5] studied the kinetics of base-catalyzed alcoholysis of soybean oil with
methanol at 6:1 M ratio of alcohol to oil. They showed that pseudo
rst order kinetics provided a satisfactory mechanism with experimental results at a large molar excess of alcohol and second order
kinetics provided a satisfactory mechanism with experimental results at 6:1 M ratio of alcohol to oil. Since then, there were several
kinetic studies of vegetable oil alcoholysis have been reported [6
8]. Most of them reported that the best kinetics mechanism of vegetable oils alcoholysis appeared to be second order. However, most
of those second order reaction models did not consider all possible
side reactions that occur during the alcoholysis. The real vegetable
oils contain not only glycerides (triglyceride, monoglyceride and
diglycride) but it also contain a considerably amount of water, FFAs
and some other impurities. Due to these impurities and the side
reactions, the nal composition of alcoholysis reaction was not
predicted accurately and completely by the second order kinetic
model. Requirement for a more complete kinetic reaction model
that provide nal concentration estimation accurately is very useful for biodiesel production process designers to achieve international standard biodiesel product quality with reasonable
production cost.

0016-2361/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fuel.2010.01.017

Please cite this article in press as: Berchmans HJ et al. Kinetic study of hydroxide-catalyzed methanolysis of Jatropha curcaswaste food oil mixture for
biodiesel production. Fuel (2010), doi:10.1016/j.fuel.2010.01.017

ARTICLE IN PRESS
2

H.J. Berchmans et al. / Fuel xxx (2010) xxxxxx

In 2002, Komers et al. [9] derived a kinetic model from proposed

mechanisms for all competing reactions that take place during
alcoholysis. The reactions include alkoxide formation, alcoholysis,
and saponication. Their kinetic model is the only attempt to formulate chemical kinetics reaction model of vegetable oil alcoholysis completely. The model assumed that the vegetable oil was
purely triglyceride, FFAs concentration on the vegetable oil was
negligible and the saponication of FFAs did not occur during alcoholysis. Review and reanalysis of Komers kinetic model has been
carried out by Tuner [10] by extending the kinetic model and
involving FFAs saponication to the developed kinetic reaction
model. However, Tuner work focused only on theoretical aspects
and examination of previous research and experimental data without implementing any real experiment to verify his developed kinetic reaction model.
The purpose of this work is to present experimental and kinetic
study of methanolysis of J. curcaswaste food oil mixture using
potassium hydroxide as a catalyst. The kinetic study introduces
enhancement of Komers kinetic model by considering saponication of FFAs in J. curcaswaste food oil mixture methanolysis at any
time under certain reaction conditions.
2. Material and methods
2.1. Materials
J. curcas oil in this experimental study was supplied from Energy Technology Laboratory, BPP Teknologi, Kawasan Puspiptek
Serpong, Indonesia. Waste food oil was obtained from Gunma prefecture area in Japan. The FFAs content in the oils was 5.53%wt for J.
curcas oil and 0.45%wt for waste food oil. The mass ratio of J. curcas
oil to waste food oil within the oils mixture is 1:9. At this mass ratio, the FFAs level of the mixture was about 1%wt. Fatty acids composition of the oils is given in Table 1. Methyl Heptadecanoate as
Fatty Acid Methyl Ester (FAME) analytical internal standard was
obtained from SigmaAldrich, Tokyo, Japan. FAME standards, certied methanol of 99.8% purity, 99.0% heptane and potassium
hydroxide of 85% purity, and other chemicals were purchased from
Wako Chemicals, Tokyo, Japan. Glycerin and glycerides test standard kit was purchased from SigmaAldrich, Tokyo, Japan.
2.2. Apparatus
Experiments were carried out in a 500 mL necked batch reactor,
where the total volume of reactive was 224 mL. The reactor was
sealed tightly with a silicon rubber cap to retain any outow of
vaporized mixture and was equipped with a mechanical stirrer,
digital thermometer and sampling port. The reactor was immersed

in a constant-temperature water bath. Inside the glass water bath,

there was an electric heater (100 V, 500 W) and a shielded K type
thermocouple probe. A voltage regulator controls the temperature
level and heat supply of water bath.
2.3. Procedures
The batch reactor initially was charged with 150 g of J. curcas
waste food oil mixture. The batch reactor was then heated for
2 h at atmospheric pressure at various temperatures (40, 50 and
60 C) and at various mixing speed (400, 600, 900, 1000 and
1200 rpm). At the very early of reaction time, a potassium methoxide solution, which is solution of a 6:1 M ratio of methanol to oil
and a 1.0% mass ratio of KOH to oil, was added to the batch reactor
immediately. In the kinetics model reaction analysis, the experiments were carried out on the same initial composition reaction
at xed temperature 50 C and at mixing speed 900 rpm to reduce
the effect of mass transport on the kinetic reaction. The mechanical
stirrer was started immediately as the reaction was timed. After
2 h of reaction, the mixture was allowed to settle down for overnight before separating the glycerol. The top layer that contain esters fraction was removed in a separated bottles and weighted. For
biodiesel purity analysis the esters were washed with hot water
several times to remove impurities. The water in the biodiesel
was removed by chemical drying agent and by heating the biodiesel product in a furnace at 107 C with 200 ml/min nitrogen ow
for 1 h.
2.4. Sampling and analysis
Sampling and analysis of the reaction mixture in this section
were required to determine the reaction mixture compositions
and concentration at any reaction time. Reaction mixture samples
of about 13 mL were withdrawn from the batch reactor at prespecied time (0.5, 1, 2, 3, 4, 5, 6, 8, 10, 12, 15, 25, 30, 40, 90 and
120 min). At the beginning of the reaction, the samples were taken
more frequently. For glycerides, ester (E) and fatty acids salt or
soap (A) analysis, each sample was collected in 3.5 mL glass vial
and was quenched immediately with 1 mL, 0.5 N of hydrochloric
acid (HCl) and cooled down to 0 C immediately in an ice batch..
All the samples mixture then was centrifuged at 3000 rpm for
30 min. Two layers were formed after centrifugation. The top layer
was Ester Phase (EP) that contained triglyceride (TG), diglyceride
(DG), monoglyceride (MG), E, FFAs (FFA) and A. The bottom layer
was Glycerol phase (GP) that contains glycerol (G), water (W)
and methanol (M). The EP of the mixture in the samples was separated and dried. In glycerides analysis, 100 mg of the EP was sylated with N-methyl-N-trimethylsilyltriuoracetamide (MTSFA);

Table 1
Fatty acid composition of Jatropha curcas oil and waste food oil.
Fatty acid

Formula

Lauric
C12H24O2
Myristic
C14H28O2
Palmitic
C16H32O2
Palmitoleic
C16H32O2
Stearic
C18H36O2
Oleic
C18H34O2
Linoleic
C18H32O2
Linolenic
C18H30O2
Arachidic
C20H40O2
Behenic
C22H44O2
Total saturated fatty acid
Total monounsaturated fatty acid
Total polyunsaturated fatty acid
a

Structurea

Jatropha curcas oil (wt.%)

Waste food oil (wt.%)

12:0
14:0
16:0
16:1
18:0
18:1
18:2
18:3
20:0
22:0

0.05
14.70
0.65
6.75
40.05
36.60
0.95
0.15
0.10
21.50
40.70
37.80

0.20
11.60
0.40
2.40
44.30
34.30
6.80

14.20
44.70
41.10

xx: y indicates xx carbons in the fatty acid chain with y double bonds.

Please cite this article in press as: Berchmans HJ et al. Kinetic study of hydroxide-catalyzed methanolysis of Jatropha curcaswaste food oil mixture for
biodiesel production. Fuel (2010), doi:10.1016/j.fuel.2010.01.017

ARTICLE IN PRESS
3

H.J. Berchmans et al. / Fuel xxx (2010) xxxxxx

added with internal standard; and diluted with heptane. Then, it

was analyzed by Gas Chromatograph (GC) GC-2010 Shimadzu that
equipped with a ame ionization detector and a capillary column
of 95% dimethylpolysiloxane (ZB-5HT Inferno, 15 m  0.32 mm 
0.10 lm). The GC oven was initiated at 50 C for 1 min, heated at
rate 15 C/min up to 180 C, at rate 7 C/min up to 230 C, at rate
10 C/min up to 380 C, and then hold it at 380 C for 5 min. The
analysis of the sample solution was carried out by injecting 1 ll
of the sample solution into the GC. The formed glycerides were
identied by comparing its retention time to the retention time
of standard glycerides derivative. Soap analysis was determined
by potentiometric titration method in which 1 ml of EP was titrated by using 0.01 N of KOH and analyzed by using Automatic
Titrator TOA-DKK, Model AUT-701. The quantity of ester in the
EP was also analyzed using a Gas Chromatograph. International
standard method EN 14214 was used to examine the concentration
of TG, DG, MG, glycerol, E in the EP. The W concentration was
determined by dilution of 1 mL of actual reaction mixture sample
in waterless chloroform. Then, the Karl Fischer method was used
to analyze the W content. Determination of basicity (ion OH) in actual reaction mixture was described in detail by Komers [11].
In this work, the initial composition of the J. curcaswaste food
oil mixture is TG 89.9%wt, DG 6.6%wt, MG 2.4%wt, FFA 1.0%wt and
W 0.1%wt. The experimental mole concentration of glycerides (TG,
DG, and MG), E, and G were calibrated by using the experimental
weight percentages of reaction compositions, their mean molecular weights, and the mechanism of stepwise reactions. The calibration was based on the mass conservation of reaction mixture and
the mole conservation of glycerol molecule. The mean molecular
weights of TG, DG, MG, FFA, E, and A 857.4, 602.3, 347.2, 278.5,
291.1, and 316.6 kg kmol-1 respectively. Those mean molecular
weights were calculated from the known fatty acids composition
of J. curcas oil and waste food oil in Table 1.
2.5. Kinetic modeling
The overall reaction scheme for methanolysis of triglyceride is
presented in Fig. 1. The stoichiometry of reaction require 3 mol
of M and 1 mol of TG to produce 3 mol of fatty acid methyl esters
(E) and 1 mol of G. A kinetic reaction model that had been studied
by Freedman [4] revealed that the main methanolysis reaction of
TG in the presence of catalyst were consecutive and competitive
reversible reactions which yields DG and MG as the intermediates.
Consequently, there were 3 stepwise reversible reactions in methanolysis. However, Komers et al. [9] argued that beside the methanolysis reaction, all present substances might take part in various
different reactions with the methoxide solution of KOH. The dominant side reactions were the saponication of TG, DG, MG, and E.
The development of the chemical kinetic model was proceeded by
simplifying assumptions that of all the theoretically possible reactions only the alcoholysis of glycerides and the saponication of
TG, DG, MG, and E, took place to form products; all of the isomers
of TG, DG, MG, and E proceeded at the same rate with the same
mechanism; and methanolysis reactions were catalyzed by methoxide ions RO (RO) as a true catalyst. Formation of methoxide ions
RO was reversible reaction and produced W as shown in Eq. (1).

CH3-O-CO-R1

CH2-O-CO-R1
CH-O-CO-R2

3CH3OH

CH2-O-CO-R3
Triglyceride (TG)

Methanol (M)

CH3-O-CO-R2

CH2-OH

+ CH-OH

CH3-O-CO-R3

CH2-OH

Methyl Ester (E)

Glycerol (G)

Fig. 1. Overall scheme of the triglyceride methanolysis scheme.

M OH $ RO W

1


Concentrations of OH (OH) and RO ions were much smaller

than glycerides and M concentration. It is shown on Eq. (2) that
the methanolysis kinetic reaction scheme consists of three reversible reactions. Beside glycerides methanolysis reaction, there are
also glycerides and E saponication reaction. It is shown on Eq.
(3) that the saponication kinetic reaction scheme consists of four
irreversible reactions.
k1

TG M OH $ DG E OH
k2

k3

DG M OH $ MG E OH
k4

k5

MG M OH $ G E OH
k6

Based on these assumptions, the complete reaction scheme of

Komers kinetic model has 8 reaction components (TG, DG, MG,
G, E, M, OH and A).
k8

E OH ! A M
k9

TG OH ! A DG
k10

DG OH ! A MG
k11

MG OH ! A G
However, this kinetic reaction model requires an additional
reaction equation if it is used for study the effect of FFA content.
If participation of FFA in the kinetic reaction model has to be considered then saponication of FFA must be involved in the set of Eq.
(3). Thus, Eq. (4) is additional saponication for the set of Eq. (3).
k12

FFA OH ! A W

The saponication of FFA is an adverse side reaction because it

consumes catalyst to produce water and more soap. Therefore, as a
result, the new system of kinetic reaction model that considers FFA
saponication consist 10 reaction components (TG, DG, MG, G, E,
M, A, OH, FFA and W).
The reaction rate expression of new kinetic reaction model is
developed based on the chemical kinetic reaction from Eqs. (2)
(4) above. By assuming that all involved reactions are elementary,
steady state, no signicant volume change and no polarity change
during the reaction then based on the law of mass action theory,
the differential form of rate expression for each reaction component is as follows:

dTG
OHk1 TGM k2 DGE  k9 TG
dt
dDG
OHk1 TGM  k2 DGE  k3 DGM k4 MGE
dt
k9 TG  k10 DG
dMG
OHk3 DGM  k4 MGE  k5 MGM k6 GE
dt
k10 DG  k11 MG
dG
OHk5 MGM  k6 GE k11 MG
dt
dE
dM

OHk1 TGM  k2 DGE k3 DGM
dt
dt
 k4 MGE k5 MGM  k6 GE  k8 E
dOH
dA

OHk8 E  k9 TG  k10 DG  k11 MG  k12 FFA
dt
dt
dW
dFFA
5

k12 FFAOH
dt
dt
In the Eq. (5), k1 to k12 are reaction rate constants (L mol1 s1);
[TG], [DG], [MG], [G], [M], [E], [OH], [A], [FFA] and [W] are mole
concentrations (mol L1) of TG, DG, MG, G, M, E, OH, A, FFA and W.

Please cite this article in press as: Berchmans HJ et al. Kinetic study of hydroxide-catalyzed methanolysis of Jatropha curcaswaste food oil mixture for
biodiesel production. Fuel (2010), doi:10.1016/j.fuel.2010.01.017

ARTICLE IN PRESS
4

H.J. Berchmans et al. / Fuel xxx (2010) xxxxxx

100

2.6. Solving and evaluation of kinetics reaction model

jyi texp  yi tcal j=yi texp

3. Results and discussion

3.1. Effect of mixing intensity and reaction temperature
A number of methanolysis reaction experiments were carried
out by varying the mixing speed from 400 to 1200 rpm, varying
the reaction temperature from 50 to 60 C, using molar ratio of
methanol to oil 6:1 and KOH concentration 1%wt of oil. The purpose of these experiments was to determine a suitable reaction
condition for the following kinetic reaction model experiment.
Fig. 2 shows the rate conversion of J. curcaswaste food oil mixture.
This gure reveals that if mixing speed was insufcient, as in the
case 400 rpm, then the rate of oil conversion was very low. It converts the oil mixture to 65.8%wt of esters within 60 min. Some of
researchers [5,7] observed that mixing speed effect was the most
signicant during the early stage the reaction (e.g. in the rst
10 min). The mixture of oilmethanol behaved as heterogeneous
phase in the rst few minutes of the reaction. Thus, in this early
stage the reaction controlled by the mass transfer. When esters
formed, it dissolved in oil and methanol. As a single phase mixture
was established during methanolysis process, the mixing effect became insignicant and the reaction rate was primarily inuenced
by the reaction temperature. If purely chemical kinetic reaction

Table 2
Values the kinetic rate constants.
Rate constant

Values (mol1 L s1)

k1
k2
k3
k4
k5
k6
k8
k9
k10
k11
k12

2.890  102
6.899  105
4.241  102
6.122  109
6.179  102
9.473  109
2.814  103
2.814  103
1.205  1010
2.814  103
1.680  101

80
70
60
1200 rpm

50

1000 rpm

40

900 rpm

30

700 rpm

20

600 rpm
400 rpm

10
0

20

40

60

80

100

Reaction time (min)

Fig. 2. Effect of mixing speed on conversion of Jatropha curcaswaste food oil
mixture methanolysis at 50 C using 6:1 methanol to oil molar ratio with KOH
concentration 1%wt of oil.

100
90
Oil conversion ( %wt)

RAD % 100=n

m X
t
X

Oil conversion (%wt)

90

A computer code of RungeKutta 4th order for differential

equations of reaction rate of Eq. (5) was developed in visual basic
of Excell. Mole concentrations of each reaction component with respect to time were calculated from the experimental data. KINTECTUS v3.961 was employed as a curve tting and optimization
program. This simulation program can be used freely for academic/educational purpose [12]. Its use is very simple because it
requires only input of the system of stoichiometric equations
describing the studied model, initial concentrations and concentrations with respect to time for all reaction components and the estimation of the rate constants into a spreadsheet of Excell. The rate
constants k1 to k12, which are shown in Table 2, were obtained by
curve tting the experimental data. The correctness of the studied
model is quantied through correlation coefcient (R2) and Relative Average Difference (RAD). The formula of RAD is given on
Eq. (7). In this equation, n is the number of measurements in interval (0: t), m is the number reactants, yi(t) is concentration of component i in time t, exp and cal denote experimental and calculated
value. Hence, RAD indicates the average relative difference in % between all experimental and the relevant calculated concentrations
of all determined species in the given experiment.

80
70
60
50
40

60C

30

50C

20

40C

10
0
0

20

40
60
80
Reaction time (min)

100

120

Fig. 3. Effect of the temperature on conversion of Jatropha curcaswaste food oil

mixture methanolysis at 900 rpm using 6:1 methanol to oil molar ratio with KOH
concentration 1%wt of oil.

model was used to predict the behavior of methanolysis reaction,

then the effect of mass transport in the reaction rate of methanolysis must be reduced or decreased by high mixing speed. At mixing
speed 900 rpm, the oil conversion reached 96%wt within less than
20 min. At the highest mixing speeds 1200 rpm, the oil conversion
increased sharply to 90%wt within less than 10 min. Fig. 3 shows
that methanolysis reaction at 900 rpm clearly was controlled by
temperatures only. At temperature 40 C, the conversion rate of
oil started with a sudden surge. When the temperature was increased from 40 C to 60 C, the conversion rate of oil conversion
also increased. In all reaction temperatures, the conversion rate
of oil was very high during the early stage of the reaction, which
was an indication of mass transfer effect reduction (at 900 rpm
of mixing speed). As the reaction time increased, the rate oil conversion was declined sharply because the methanolysis reaction
moved towards equilibrium state condition.
3.2. Dependence of concentration to the reaction time
Dependence of molar concentration of glycerides, G, M and E to
the reaction time for the simulation and experimental results are
shown in the prole of Fig. 4. Several other kinetic studies showed
similar typical curve prole [6,7]. Figs. 5 and 6 show simulation
and experimental results of the reaction mixture concentration
during the early stage of reaction (the rst 15 min of reaction
time). In the rst 2 min of reaction time, the TG concentration decreased sharply as the sudden increase of E concentration. The

Please cite this article in press as: Berchmans HJ et al. Kinetic study of hydroxide-catalyzed methanolysis of Jatropha curcaswaste food oil mixture for
biodiesel production. Fuel (2010), doi:10.1016/j.fuel.2010.01.017

ARTICLE IN PRESS
H.J. Berchmans et al. / Fuel xxx (2010) xxxxxx

Fig. 4. Simulation of kinetic modeling curves and experimental data points for the
concentration of reaction mixture of Jatropha curcas-waste food oil mixture
methanolysis at 900 rpm, temperature 50 C using 6:1 methanol to oil molar ratio
with KOH concentration 1%wt of oil. (Solid lines, simulation curve; h, triglyceride;
s, diglycride; 4, monoglyceride; , glycerol; N, methyl ester; +, methanol).

highest concentrations of DG and MG observed were visible at the

rst 3 min of reaction time. Their levels then decreased when reaction time increased. Concentration of FFA, W, OH and A are also depicted in Fig. 7. The kinetic reaction model simulation and
experimental results show that ion OH concentration decrease before the end of reaction time (120 min). When OH concentration
decreased the A production increased sharply. The decreasing of
ion OH concentration and increasing of A concentration conform
the saponication side reaction hypothesis of our kinetic reaction
model. The undesired saponication of glycerides and E consumed
the ion OH, and the present of FFA in the reaction mixture accelerated the consummation process. The calculated rate constant k12 of
FFA saponication reaction in Table 2 shows that FFA saponication rate is 60 times larger than saponication rate of glycerides
and E, and also 4 times larger than forward reaction rate of glycerides methanolysis. Clearly, there were competitions between
methanolysis and saponication reactions. The chemical properties of the true catalyst (methoxide ion) are the key role on both
reactions because the mechanism reactions of both reactions involve the attack of catalyst [1315]. Thus, methoxide ion as a catalyst in this methanolysis is very sensitive to saponication
reaction of FFA. In our experiment, 1%wt FFA content in J. curcas
waste food oil mixture had no distinct negative effect to the methanolysis kinetic reaction. However, if FFA content in the oil mixture
was increased to more than 1%wt (by increasing the mass ration of

Fig. 5. The rst 15 min of reaction time of kinetic modeling simulation curves and
experimental data points for the concentration of reaction mixture of Jatropha
curcaswaste food oil mixture methanolysis at 900 rpm, temperature 50 C using
6:1 methanol to oil molar ratio with KOH concentration 1%wt of oil. (Solid lines,
simulation curve; h, triglyceride; , glycerol; N, methyl ester).

Fig. 6. The rst 15 min of reaction time of kinetic modeling curves simulation and
experimental data points for the concentration of reaction mixture of Jatropha
curcaswaste food oil mixture methanolysis at 900 rpm, temperature 50 C using
6:1 methanol to oil molar ratio with KOH concentration 1%wt of oil. (Solid and
dotted lines, simulation curve; s, diglycride; 4, monoglyceride).

Fig. 7. The rst 15 min of simulation of kinetic modeling curves and experimental
data points for the concentration of OH, A, FFA and W. (Lines, simulation curve; s,
OH; 4, A; h, FFA; N, W).

J. curcas oil in the oil mixture), then it denitely would accelerate

FFA saponication reaction and consume much more catalyst until; at certain level it would disturb the completeness of methanolysis reaction; in the worst case it would stop methanolysis reaction
immediately. Figs. 47 show that the species concentration from
simulation result on the kinetic model gives an acceptable t to

Fig. 8. Methyl ester purity of the experimental data points and simulation curve of
Jatropha curcaswaste food oil mixture methanolysis at 900 rpm, temperature 50 C
using 6:1 methanol to oil molar ratio with KOH concentration 1%wt of oil. (Line,
simulation curve; h, experimental data points).

Please cite this article in press as: Berchmans HJ et al. Kinetic study of hydroxide-catalyzed methanolysis of Jatropha curcaswaste food oil mixture for
biodiesel production. Fuel (2010), doi:10.1016/j.fuel.2010.01.017

ARTICLE IN PRESS
6

H.J. Berchmans et al. / Fuel xxx (2010) xxxxxx

Table 3
Comparison of end concentrations and composition of the reaction mixture.
Time (min)

Concentration (mol L1)

DG

MG

OH

FFA

New kinetic reaction model

120
0.0100
15
0.0306

TG

0.0082
0.0117

0.0101
0.0218

0.7939
0.7989

0.0007
0.0

0.0006
0.0006

0.2449
0.2449

2.1232
2.0801

0.1195
0.1188

Second order kinetic reaction model

120
0.0114
15
0.0294

0.0093
0.0145

0.0113
0.0129

0.7929
0.7661

2.2405
2.2574

Experimental data
120
15

0.0096
0.0379

0.0074
0.0143

0.0091
0.0252

0.7999
0.7946

0.0
0.0

0.2460
0.2459

2.1458
2.0875

0.100
0.100

0.0020 (max)

0.0029 (max)

0.0198 (max)

0.0019 (max)

EN 14214 & ASTM D6854

the experimental data. The calculated values of forward rate constants k1, k3, and k5 show a consistency because it values higher
than the values of backward rate constant k2, k4, and k6. It indicates
that the forward reactions are much faster than reverse reactions.
It means that the glycerides methanolysis reactions lay toward the
esters production. Therefore, the proposed kinetic reaction model
with the obtained rate constant provides a satisfactory mechanism
for methanolysis of J. curcaswaste food oil that contains FFA.
3.3. Effect of reaction time on purity of biodiesel
Biodiesel purity is dened as the weight percentage of esters
content in the biodiesel product. High rate of oil conversion during
the early stage of methanolysis reaction affect the biodiesel purity.
Fig. 8 shows experimental data points and simulation result of biodiesel purity on the methanolysis reaction for the rst 15 min. Simulation result shows that kinetic reaction model and calculated
kinetic data were able to predict the biodiesel purity experiment
data under given reaction conditions. In the early stage of reaction
time, the purity increased rapidly. At 15 min of the rst reaction
time, the biodiesel purity was 92.7%wt, which is below the acceptable limit of standard biodiesel purity. However, experiment data
and simulation result showed that increasing the reaction time
to 1 h and more had increased the biodiesel purity to more than
97.1%wt. According to the European Union standards (EN 14214)
for alternative diesel fuels, the minimum acceptable limit of biodiesel purity is 96.5%wt.
3.4. End concentration and correctness of analysis
Komers et al. [9] studied the kinetics and mechanism of the
KOH-catalyzed methanolysis of Rapeseed oil at 2:16:1 M ratios
of methanol to Rapeseed oil and at the molar ratio of KOH to Rapeseed oil 0.24. They claim that the probability of validity of their kinetic reaction model was 78% agreement with the experimental
data. Their conclusion was based on the RAD calculation. RAD
was the objective criteria of the correctness of the kinetic reaction
model at given condition (initial reaction mixture composition and
reaction condition) and experimental data. Thus, by similar method we verify the correctness of our kinetic reaction model. Moreover, we also verify the potency of proposed kinetic reaction
model for practical application. It is necessary to have more complete kinetic model that capable to describe the reaction mechanism of vegetable oil methanolysis and saponication. However,
the proposed kinetic reaction model should also be useful to estimate the reaction mixture composition at end of reaction time
accurately and completely. Due to that reason, we compare our
new kinetic reaction model with second order kinetic reaction

Table 4
Comparison of relative average difference and coefcient correlation (R).
Kinetic model

RAD (%)
(0120) min

RAD (%)
(15120) min

New kinetic reaction model

Second order kinetic reaction model

2.403
3.999

1.407
5.001

0.894
0.895

model of previous kinetic studies [6,7]. The rate expression of the

second order kinetic model involved only methanolysis reaction
of glycerides. By using the same experimental data in the rate constant calculation, comparison of two kinetic models is focused on
the calculation of end concentration and composition of the reaction mixture. The analysis of end concentration and composition
are very important when estimation or prediction the yield and
quality of biodiesel product are required. The information is required to estimate the effort size of rening stage. Lower quality
of biodiesel, which is indicated from its nal concentration and
composition, requires larger effort or larger cost in rening stage
process. The information is also required to adjust the reaction
condition and the initial reaction mixture composition, because
methanolysis reaction involving vegetable oil that has variation organic compound mixture composition from every different feedstock. Therefore, a better kinetic reaction model should be able
to estimate end concentration and composition of reaction mixture
accurately and completely. Table 3 shows the comparison of concentrations end and composition of the proposed kinetic model
simulation results, second order kinetic model simulation results
and the experimental data. The new kinetic model end concentrations simulation result at the rst 12 min reaction time and at the
end of 120 min reaction time shows better estimation to the experimental data. In addition, the new kinetic model was able to predict end concentrations of OH, A, FFA and W, which is very
useful for standard quality assessment of biodiesel product. Goodness t of the new kinetic reaction model to the experimental data
was quantied by the values of RAD and correlation coefcient as
shown in Table 4. RAD of new kinetic model at any reaction time is
smaller than RAD of the second order kinetic reaction model. It
means that the proposed kinetic reaction model estimation result
has smaller deviation from the experimental data.

4. Conclusion
The experiment and simulation of hydroxide-catalyzed methanolysis of J. curcaswaste food oil mixture is presented. The proposed kinetic model provides better estimation for the
hydroxide-catalyzed methanolysis reaction of J. curcaswaste food

Please cite this article in press as: Berchmans HJ et al. Kinetic study of hydroxide-catalyzed methanolysis of Jatropha curcaswaste food oil mixture for
biodiesel production. Fuel (2010), doi:10.1016/j.fuel.2010.01.017

ARTICLE IN PRESS
H.J. Berchmans et al. / Fuel xxx (2010) xxxxxx

oil mixture under given condition. The model is able to estimate

the effect of FFA saponication and has described that the presence
of FFA has negative effect on the hydroxide-catalyzed methanolysis. Although verication the potency of the kinetic reaction model
requires more data from experiments, the implementation of the
model provides useful information for future works.
Acknowledgements
This work was supported by Department of Chemical and Environmental Engineering, Gunma University, Kiryu, Japan. The
authors also express their gratitude to Dr. Youichi Kamiishi from
Science and Technology Promotion Ofce, Gunma Perfecture, Japan
for his support and kind assistance.
References
[1] Leung DYC, Guo Y. Transesterication of neat and used frying oil: optimization
for biodiesel production. Fuel Process Technol 2006;87(8):8390.
[2] Berchmans HJ, Hirata S. Biodiesel production from crude Jatropha curcas L. seed
oil with a high content of free fatty acids. Bioresour Technol 2008;99:171621.
[3] Achten WMJ, Verchot L, Franken YJ, Mathijs E, Singh VP, Aerts R, et al. Jatropha
bio-diesel production and use. Biomass Bioenergy 2008;32:106384.
[4] Freedman B, Pryde EH, Mounts TL. Variables affecting the yields of fatty esters
from transesteried vegetable oils. J Am Oil Chem Soc 1984;61(10):163843.

[5] Noureddini H, Zhu D. Kinetics of transesterication of soybean oil. J Am Oil

Chem Soc 1997;74(11):145763.
[6] Vicente G, Martnez M, Aracil J, Esteban A. Kinetics of sunower oil
methanolysis. Ind Eng Chem Res 2005;44:544754.
[7] Bambase Jr ME, Nakamura N, Tanaka J, Matsumura M. Kinetics of hydroxidecatalyzed methanolysis of crude sunower oil further production of fuel-grade
methyl esters. J Chem Technol Biotechnol 2007;82:27380.
[8] Darnoko D, Cheryan M. Kinetics of palm oil transesterication in a batch
reactor. J Am Oil Chem Soc 2000;77(12):12637.
[9] Komers K, Skopal F, Stloukal R, Machek J. Kinetics and mechanism of the KOHcatalyzed methanolysis of Rapeseed oil for biodiesel production. Eur J Lipid Sci
Technol 2002;104(11):72837.
[10] Tuner TL. Modeling and simulation of reaction kinetics for biodiesel
productions. Master degree thesis. USA: North Carolina State University; 2005.
[11] Komers K, Skopal F, Machek J, Stloukal R. Biodiesel from rapeseed oil, methanol
and KOH3. Analysis of composition of actual reaction mixture. Eur J Lipid Sci
Technol 2001;103:36371.
[12] Ianni JC. A comparison of the BaderDeuhard and the CashKarp Runge
Kutta integrators for the GRI-MECH 3.0 model based on the chemical kinetics
code KINTECTUS, computational uid and solid mechanics. Oxford, UK:
Elsevier Science Ltd.; 2003. p. 136872.
[13] Doell R, Konar SK, Boocock DGB. Kinetic parameters of a homogeneous
transmethylation of Soybean oil. J Am Oil Chem Soc 2008;85:2716.
[14] Komers K, Skopal F, Stloukal R. Determination of the neutralization number for
biodiesel fuel production. Fett/Lipid 1999;99(2):524.
[15] Komers K, Machek J, Stloukal R. Biodiesel from Rapeseed oil, methanol and
KOH2. Composition of solution of KOH in methanol as reaction partner of oil.
Eur J Lipid Sci Technol 2001;103:35962.

Please cite this article in press as: Berchmans HJ et al. Kinetic study of hydroxide-catalyzed methanolysis of Jatropha curcaswaste food oil mixture for
biodiesel production. Fuel (2010), doi:10.1016/j.fuel.2010.01.017