Fuel Processing Technology 125 (2014) 34–40

Contents lists available at ScienceDirect

Fuel Processing Technology

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

Kinetics of reactive extraction/in situ transesterification of rapeseed oil

Rabitah Zakaria a,⁎, Adam P. Harvey b
a
Department of Process and Food Engineering, Faculty of Engineering, University Putra Malaysia, UPM Serdang, 43400 Selangor, Malaysia
b
School of Chemical Engineering and Advanced Materials, Newcastle University, NE1 7RU, UK

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

Article history: The kinetics of “reactive extraction” or “in situ transesterification” of rapeseed with methanol to produce biodie-
Received 30 August 2012 sel were investigated. It is hypothesised that in situ transesterification occurs through the reaction of triglyceride
Received in revised form 25 September 2013 and methanol inside the seed particles followed by diffusion of mono and diglycerides, esters and glycerol into
Accepted 16 March 2014
the bulk solvent. A model was developed based on this reaction/extraction mechanism and was found to be gen-
Available online 6 April 2014
erally consistent with experimental results. The effective diffusion coefficient of esters in methanol was found to
Keywords:
be 3.5 × 10−12 m2/s. The model reveals that at catalyst concentrations below 0.1 mol/kg-solvent, the reactive ex-
Biodiesel traction rate was controlled by the reaction of triglycerides to esters. However, at higher catalyst concentrations
Rapeseed (N0.1 mol/kg-solvent), the process was controlled by the rate of diffusion of the products. The diffusion coeffi-
Canola cient increases linearly with increasing temperature and the temperature dependence of the reaction rate con-
Kinetics stants can be described by an Arrhenius relationship. External mass transfer does not influence the extraction
Reactive extraction rate at the conditions used in these experiments.
In situ transesterification © 2014 Elsevier B.V. All rights reserved.

Nomenclature 1. Introduction

Ap particle surface area m2 Biodiesel is a fuel derived from renewable resources, in practice
CA concentration of species A inside the particle mol/L largely edible and inedible oil-bearing seeds, palm fruits and waste
CA,b concentration of species A in the bulk solvent mol/L cooking oil. Biodiesel can be considered as a versatile fuel making it an
CA,s concentration of species A at the particle surface mol/L
dp average particle diameter m
attractive alternative to petroleum diesel. It is most commonly pro-
Deff effective diffusivity m2/s duced by the reaction of vegetable oils with an alcohol to produce
D diffusion coefficient m2/s fatty acid alkyl ester. The fuel can be readily used in most diesel engines
Ea activation energy J/mol without any modification and engine performance has been shown to
k reaction rate constant L/mol/s
be comparable to that of conventional diesel fuel [1]. The use of biodiesel
km,A external mass transfer coefficient of species A m/s
ko Arrhenius constant L/mol/s reduces the emission of carbon monoxides, particulates and hydro-
nA moles of species A in the particle mol carbon from the engine [1]. Furthermore, a biodiesel blend has lubri-
Np number of particle cating properties that can reduce engine wear and extends engine
r radius inside seed particle m life [2,3].
r′A reaction rates of species A mol/L/s
In the conventional biodiesel process, the first step is the extraction of
r′A,b reaction rates of species A in the bulk solvent mol/L/s
rm mean radius between r and delta r m the oil from the plant or animal sources followed by transesterification
RO outer radius of seed particle m with an alcohol to produce fatty acid alkyl ester. Extraction of the oil is
R universal gas constant commonly performed using solvents such as hexane or dichloromethane,
Re Reynolds number
which have high affinities for triglycerides. Mechanical pressing is also
Sh Sherwood number
Sc Schmidt number commonly used to extract the oil although it often yields less oil than sol-
t time s vent extraction resulting in a higher residual oil in the waste biomass.
T temperature K Once extracted, the oil may need to be purified before entering the
Vb solvent volume m3 transesterification unit. However, it has been demonstrated that the ex-
WA flux of species A mol/m2
traction and transesterification process can be performed simultaneously
ε porosity
using a single alcohol, hence eliminating a separate extraction step and
avoiding the use of solvents such as hexane [4]. This combined route,
⁎ Corresponding author. Tel.: +60 389464301; fax: +60 389464440.
which is often termed “reactive extraction” or “in situ transesterification”,
E-mail addresses: rabitah@upm.edu.my (R. Zakaria), adam.harvey@ncl.ac.uk has the advantages of simplifying the biodiesel production process as well
(A.P. Harvey). as potentially reducing production cost.

http://dx.doi.org/10.1016/j.fuproc.2014.03.019
0378-3820/© 2014 Elsevier B.V. All rights reserved.
 R. Zakaria, A.P. Harvey / Fuel Processing Technology 125 (2014) 34–40
In reactive extraction or in situ transesterification, the biomass is di-
rectly reacted with the alcohol and a catalyst. The resulting extracts are
mainly biodiesel and glycerol. The alcohol can be various short-chain al-
cohols such as methanol, ethanol or butanol, although methanol is com-
monly used due to its lower price and because it reacts the most rapidly.
Methanol, however, is a poor solvent for triglycerides and is unable to
extract most types of triglycerides to any significant extent [5,6]. How-
ever, when it is used in in situ transesterification, the oil is extracted
directly as fatty acid methyl ester (FAME) [4]. The yield of FAME none-
theless greatly depends on the process parameters, such as catalyst con-
centration and methanol to oil molar ratio [7]. Since methanol is a poor
solvent for triglycerides, the mechanism that allows a significant ex-
traction of triglycerides and a subsequent transesterification to ester
to take place during reactive extraction has not been definitively
identified. Kildiran et al. (1996) suggested that the extraction of tri-
glycerides is possible due to the stepwise dissolution and subsequent
transesterification of the oil [8]. Haas et al. (2004) and Qian et al.
(2008) proposed that the alkaline catalyst could destroy the cell
walls and intracellular compartmentalization resulting in cellular
solubilisation and subsequent transesterification of triglycerides [9,
10]. On the other hand, the results from our previous studies show
that most rapeseed cell walls of in situ transesterified seeds remain
fully intact [11], and a shrinking core of oleaginous material is clearly
visible. These findings strongly suggest that triglycerides reacted
with the alkaline methanol in situ through a reaction between the
diffused methanol and the small oil bodies inside the plant cell walls.
Once the triglyceride has reacted, the reaction products (glycerol,
FAME, mono- and diglycerides) which are soluble in methanol are
able to diffuse from the seed internal into the bulk solvent. In this
paper, a fundamental model is developed based on the suggested mech-
anism in order to simulate the reactive extraction process. The model-
ling of the process mechanism is important in predicting the effects of
various process parameters and in providing insight into possible fur-
ther process improvements.
2. Materials and methods
2.1. Reagents and material
The rapeseed was kindly provided by a local farm near Newcastle
upon Tyne. The methanol used was of 99.97% purity. The solvent was
purchased from Fisher Scientific. Methyl heptadecanoate and tricaprin
internal standards were obtained from Sigma-Aldrich.
2.2. Reactive extraction
25 g of ground and sieved seed was placed in a 500 mL round-
bottom flask equipped with a condenser and an overhead stirrer and
placed in a constant temperature water bath. Alkaline alcohol, prepared
by dissolving a known amount of sodium hydroxide in methanol, was
first heated to the desired temperature using a heated circulating
water bath. Once heated, 250 ml of the alcohol was transferred to the
round-bottom flask and the reaction was carried out at 40, 50 and
60 °C until the desired reaction periods. The concentration of catalyst
was 0.1 mol/kg-solvent. The stirrer speed was maintained at 200 rpm
for all of the experiment. For the investigation on the effect of catalyst
concentration, the concentration of catalyst was varied at 0.1, 0.05 and
0.03 mol/kg-solvent while the temperature was maintained at 60 °C.
To obtain the time profile of ester and the reaction intermediates,
0.5 mL samples of the reaction mixture were taken at different time in-
tervals using a syringe and a tube from the sampling pots. 3 μL of glacial
acetic acid was added to stop the reaction. A syringe filter was used to
separate the liquid from the rest of the seed particles. A small portion
of the liquid, which contains methanol and the reaction products, was
analysed using a gas chromatograph in order to quantify the amount
of ester in the bulk methanol phase. The rest of the liquid was evaporated
35
to remove the methanol until the sample reached constant weight. The
product was then allowed to settle into two layers and the ester phase
was withdrawn from the top. The concentration of mono-, di- and tri-
glycerides was quantified using a gas chromatograph/mass spectrometer
(GC–MS).
2.3. Ester analysis
Gas chromatography analysis for the determination of individual
and total esters was based on the British Standard BS EN 14103:2003,
with some modifications. The internal standard used was methyl
heptadecanoate (MHDN). The stock solution of the internal standards
was prepared by adding 5 g of methyl heptadecanoate in 322.6 g meth-
anol. A known amount of sample (1–2 g range) was weighed accurately
into a sample vial with 1.25 g of the stock solution. The injection volume
used was 0.2 μl. The gas chromatograph used was a HP 5890 Series II,
with a Varian CP wax column with dimensions of 0.53 mm i.d., 10 μm
coating thickness and 25 m length. The carrier gas was helium at a
flow rate of 2 mL/min and the column, injector and detector tempera-
ture were maintained at 220 °C. The column separated the esters and
the methyl heptadecanoate (MHDN) into individual peaks.
The weight percent of ester in the bulk methanol phase (C) was cal-
culated using the equation below:
ð100−Area MHDN% Þ
 Solvent Ratio  mass of standard usedðg Þ
C¼ Area MHDN % :
Mass of sample usedðg Þ
2.4. Gas chromatography/mass spectroscopy (GCMS) method for glyceride
analysis
A GCMS was used to quantify the monoglycerides, diglycerides and
triglycerides in the ester phase. The method was based on BS EN
14105 with some modification to the calibration procedure, in order
to account for the higher glyceride content, which occurs particularly
during the early reaction time. Tricaprin (1,2,3-tricaproylglycerol) was
used as a basis to calibrate the glycerides.
For the analysis of monoglyceride, about 20 mg of sample was accu-
rately weighed and placed in a sample vial. 20 μl of tricaprin stock solution
(12.6 mg/mL) and 20 μL of N-methyl-N_trimethylsilyfluoraacetamide
(MSTFA) were added to the sample vial. The solution was left to silylate
for 15 min, after which 2 mL of heptane was added. For identification of
TG and DG about 20 mg of sample was accurately weighed and place in
a sample vial. 100 μL of tricaprin stock solution (0.5 mg/ml) and 20 μL
of MSTFA were added to the sample vial. The solution was left to silylate
for 15 min after which 0.5 mL of heptane was added. The GCMS was fitted
with Perkin Elmer Col-elite column (PE-5HT) of 15 m length, 0.25 i.d. and
0.1 μm film thickness. The detector type in the mass spectrometer was
electron impact positive. The flow rate of helium was 1 mL/min. The
oven temperature protocol was 50 °C held for 1 min, heated to 180 °C
at 15 °C/min, then at 7 °C/min to 230 °C and at 10 °C/min to 370 °C.
The temperature was then held at 370 °C for 10 min for a total run time
of 31.5 min. The inlet line to the MS was kept at 270 °C while the MS
source temperature was kept at 250 °C.
3. Calculation/modelling
The reactive extraction of triglycerides and methanol to produce
fatty acid methyl ester was modelled as the following steps:
1. Diffusion of methanol and catalyst through the seed external bound-
ary layer to the particle surface.
2. Diffusion of methanol and catalyst into the seed to the surface of oil
droplets.
3. Reaction of triglycerides with methanol within the seed particle.
36 R. Zakaria, A.P. Harvey / Fuel Processing Technology 125 (2014) 34–40

4. Diffusion of reaction products (monoglyceride, diglycerides, glycerol, An expression of the form of Eq. (4) was applied for all the species
and fatty acid methyl esters) out to the particle surface. involved in the extraction and reaction process namely biodiesel (B),
5. Mass transfer of the products from the particle surface to the bulk glycerol (G), monoglyceride (MG), diglyceride (DG), triglyceride (TG)
liquid. and methanol (M). The partial differential equations were changed to
6. Further reaction of the intermediates to produce FAME and glycerol ordinary differential equations using the finite difference method. The
in the bulk methanol. differential equations for all species were solved using the explicit
Runge–Kutta method in MATLAB 7.1 (Math Works, Natick, MA).
The following assumptions were used in deriving the model:
The boundary conditions were:
1. The transport of triglycerides and the reaction intermediates from
At the centre of the particle, r = 0
the plant cell out to the bulk can be described by Fick's Law of
diffusion. dCA
2. The diffusion coefficient, Deff, is constant throughout the seed and in- ¼0 ð5Þ
dr
dependent of thickness.
3. A concentration boundary layer has developed around the external At the outer surface, r = Ro, the flux can be described by
surface of the seed particles.
4. The distribution of oil in the particles is uniform. dCA  
5. The seed particles are spherical and the structure of the seed is Deff ¼ km;A CA −CA;b ð6Þ
dr
homogenous.
6. The solvent in the batch reactor is perfectly mixed.
where km,A is the external mass transfer coefficient and CA,b is the con-
7. The catalyst and methanol diffuse at the same rate.
centration of species A in the bulk solvent.
The equation used to describe the concentration profile in the bulk
3.1. Model development
liquid was derived from the mole balance of the substance in the bulk
liquid as follows:
The mole balance of species A over the particle shell thickness, Δr, as
shown in Fig. 1 is
CA;b  
Vb d ¼ km;A CA;s −CA;b εAp Np þ r A;b Vb
0

 
   dn ð7Þ
2
2
dt
WAr 4πr
− WAr 4πr
þ r A 4πrm Δr ¼ A
0 2
ð1Þ
r rþΔr dt
where CA,b is the concentration of A in the bulk solvent, CA,s is the con-
where WAr is the flux of species A at distance r, r is the radial distance, nA centration of A at the particle surface, Ap is the particle surface area,
is the mole of A, t is the time, r′A is the reaction rate of species A, CAS is Np is the number of particles, ε is the porosity, Vb is the solvent volume,
the concentration of A at the interface, and rm is the mean radius be- rA,b is the reaction rate of species A in the bulk solvent, and km,A is the
tween r and Δr. external mass transfer coefficient for species A.
Dividing both sides by 4πΔr and taking the limit as Δr approaches 0, The seed oil content was determined to be 45.9 wt.%. The mass of
Eq. (1) can then be written as the following differential equation: seed used was 25 g with an average particle size of 400 μm. The initial
  concentration of the triglycerides in the seed was calculated as the num-
WAr r2 0 2 dnA ber of moles of triglycerides divided by the volume occupied by triglyc-
−d þrA r ¼ : ð2Þ
dr dt:4πdr eride plus the volume of air space in the seed. The voidage is determined
from the porosity of the seed before extraction. As shown in an earlier
The flux, WAr, is defined according to Fick′s Law as the following: work [11] (by SEM), cell walls at the particle surface are fractured dur-
ing the grinding process, causing a substantial layer of oil to form at the
dCA particle surface. The amount of surface oil can be estimated by assuming
WAr ¼ −Deff ð3Þ
dr that only the outer cells were broken during the grinding process and
was calculated to be approximately 27%. This oil should be ‘washed’
where Deff is the effective diffusivity of species A in the seed and CA is the
into the bulk liquid at a higher rate than the oil inside the seed due to
molar concentration of species A.
the absence of internal diffusion. The fast dissolution of the ‘surface
Dividing Eq. (2) by r2 and replacing the flux, W, with Eq. (3) leads to
oil’ was also reported in So and MacDonald (1986) [12], who found
the subsequent equation:
that the ‘washing’ mass transfer coefficient at 55 °C is 14 times higher
h i than the diffusion coefficient. In this study, a simplified approach was
2
1 d −Deff ðdCA =drÞr 0 dC taken by assuming that the transesterification of the surface oil occurs
þrA ¼ A: ð4Þ
r2 dr dt in the bulk solvent in order to approximate the effect of surface oil
layer in the reactive extraction process.

3.2. Expression of reaction rates

The reaction of triglyceride to methyl esters is usually modelled as

three sequential reversible reactions as shown in Fig. 2.

Triglyceride + Methanol Diglyceride + Biodiesel

Diglyceride + Methanol Monoglyceride + Biodiesel

Monoglyceride + Methanol Glycerol + Biodiesel

Fig. 1. Spherical shell in a spherical seed particle. Fig. 2. Stepwise reactions of triglycerides to FAME.
 R. Zakaria, A.P. Harvey / Fuel Processing Technology 125 (2014) 34–40 37

In most publications on the kinetics of TG transesterification, the re- concentrations of only 1.4 wt.%, 2.4 wt.% and 1.7 wt.%, respectively in
action mechanism was found to be second order for each step of the the ester phase as previously reported [16]. After a 5 minute reaction,
reaction [13–15]. The rates of the reactions can then be described as: the concentration of ester in the extracted ester phase was about
99.6% and the mono-, di- and triglycerides were low enough to pass
r1 ¼ k1 CTG CM ð8Þ the EU biodiesel standard EN14214. In comparison, during conventional
transesterification of rapeseed oil, the concentration of ester was only
about 60% after 5 min (for 6:1 methanol to oil molar ratio, 1 wt.%
r2 ¼ k2 CDG CM ð9Þ NaOH, 600 rpm stirring rate and 65 °C) [17]. The absence of a significant
amount of triglycerides and intermediates in the extracted product
throughout the course of the reactive extraction process suggests that
r3 ¼ k3 CMG CM ð10Þ
the reaction of triglycerides and the subsequent reaction of the interme-
diates have occurred in the seed particles [16]. The small amount of
mono-, di- and triglycerides present in the extracted product early in
r−1 ¼ k−1 CDG CB ð11Þ
the reaction is probably due to the dissolution of the surface layer of oil.
Fatty acid methyl ester from rapeseed oil consists mainly of oleic,
r−2 ¼ k−2 CMG CB ð12Þ palmitic, linoleic, and linolenic acid. The time profile of the percentage
of these individual esters during reactive extraction is shown in Fig. 4.
The proportion of each ester is constant throughout the reaction, show-
r−3 ¼ k−3 CG CB : ð13Þ ing that the individual fatty acids are extracted at the same rate. Hence it
can be inferred that the difference in the length of carbon chain between
Previously reported reaction rate constants (k1 through k− 3) of palmitic and oleic acid does not have a significant effect on the overall
Noureddini and Zhu [13] were used in this model. Some values of the process rate. This is expected since transesterification occurs through
rate constants, however, were adjusted in order to better fit the exper- the attack of the methoxide ion on the carboxyl group and the carbon
imental data. The relationship of reaction rate constant (k) and temper- chain length should not influence the rate in this range. The difference
ature was described by the Arrhenius Eq. (14) below: in geometric shape of the molecules due to the different number of dou-
ble bonds (0 in palmitic, 1 in oleic, 2 in linoleic and 3 linolenic) also does
k ¼ ko e
−Ea
=RT ð14Þ
not seem to cause a substantial variation in the individual ester reactive
extraction rate.
where Ea is the activation energy, ko is the Arrhenius constant, and R is
the universal gas constant. 4.2. Kinetics parameter
The external mass transfer coefficient, km, depends on the stirring rate
and the set up of the reactors. Here, it was estimated from the Sherwood Figs. 5 and 6 show a typical fitting of the experimental results for
number using Eq. (15) ester and monoglycerides at several operating temperatures. For the
ester concentration profile, the plots generally show a close fit between
km  dp 1=2 1=3 the model and the experimental data. In the initial stage, the extraction
Sh ¼ ¼ 2 þ 0:6  Re  Sc ð15Þ rate of ester was clearly lower at lower temperature, but the difference
D
reduces with increasing time. For monoglycerides, the peaked concen-
where Re is the Reynolds number, Sc is the Schmidt number, D is the liq- tration occurred earlier with higher temperatures and the rate of its
uid diffusivity, and dp is the average particle diameter. disappearance increased with increasing temperature (Fig. 6). As the re-
action of diglycerides to monoglycerides can occur inside the particle,
4. Results and discussion the faster appearance of monoglycerides peak with increasing temper-
ature is probably due to an increased diffusion rate of monoglycerides.
4.1. Reactive extraction time profile On the other hand, the increase in forward and reverse reaction rates
of the intermediate could also result in a similar profile. The predicted
Fig. 3 shows the concentration of the extracted products in the bulk profiles of monoglycerides however shows a slight discrepancy from
solvent phase. In the initial stage of the reaction, the production rate of the experimental data. A possible reason for this is due to the inaccuracy
ester was rapid, reaching 60% of the final value in 5 min, but the rate re- in estimating the amount of surface oil of the seed particles which as-
duced thereafter. The extraction is essentially complete by 30 min. The sumed that it only comes from cells at the outer surface of the seed par-
extracted product consists of only a very small amount of triglycerides ticle. In reality, some internal cells may also be damaged causing higher
and intermediates with a maximum triglyceride, mono- and diglyceride amount of surface oil. As suggested earlier, the presence of the interme-
diates in the bulk methanol is mainly due to the dissolution/reaction of
6
Concentration (wt%)

70
Ester percentage (wt%)

5
60
4
monoglycerides 50
3 diglycerides 40
triglycerides
2 30
ester
1 20
10
0
0 10 20 30 40 50 60 70 0
Time (min) 0 10 20 30 40 50 60
Time (min)
Fig. 3. Composition of reactive extraction products in the bulk solvent. Reaction condi- palmitic oleic linoleic linolenic
tions: 60 °C, 25 g seed, 0.1 m catalyst concentration, 250 ml solvent. Note: The monoglyc-
erides marker is superimposed on the di- and triglycerides marker since its concentration Fig. 4. Percentage of individual ester over time. Reaction conditions: 60 °C, 25 g seed, 0.1 m
is close to the concentration of di- and triglycerides. catalyst concentration, 250 ml solvent.
38 R. Zakaria, A.P. Harvey / Fuel Processing Technology 125 (2014) 34–40
Ester concentration (mol/l)

0.16 4.0E-12
0.14 3.5E-12

Diffusivity (m2/s)
0.12 3.0E-12
0.1 2.5E-12
0.08 2.0E-12
0.06 1.5E-12
0.04 1.0E-12 R² = 0.9884
0.02 5.0E-13
0 0.0E+00
0 20 40 60 80 100 120 140 300 305 310 315 320 325 330 335
Time (min) Temperature (K)
30 deg C-exp data 40 deg C-exp data 60 deg C-exp data
30 deg C-model 40 deg C-model 60 deg C-model Fig. 7. Diffusivity values of ester at different temperatures.

Fig. 5. Effect of temperature on ester concentration profile. Reaction conditions: 25 g seed,

0.1 m catalyst concentration, 250 ml solvent.
the surface oil. As a result, a discrepancy in estimating the effect of sur-
face oil will result in the model predicting slightly different values from
the actual data.
The resulting diffusivity coefficients of ester in methanol/rapeseed
system and reaction rate constants are determined based on the fit of
the model and the experimental data as shown in Figs. 7 and 8 respec-
tively. The diffusivity was found to decrease almost linearly with tem-
perature in the range of study, while the rate constants closely comply
with the Arrhenius equation. Sasmaz (1996) reported an experimental
diffusivity value of 3.4 × 10− 12 (m2/s) for triglyceride in a hexane/
ground rapeseed system at 55 °C [18], which is comparable to the
diffusity value of ester found in this study (3.25 × 10−12 m2/s). Since
liquid–liquid diffusion coefficient of triglyceride in hexane (calculated
using a modified Wilke–Chang correlation [19]) was found to be similar
to liquid–liquid diffusivity of fatty acid methyl ester in methanol, the ef-
fective diffusivity of both compounds in ground rapeseed should also be
in the same range.
4.3. Effect of mass transfer and reaction rate
Using the model, the effects of independently varying the diffusion
coefficient, the external mass transfer coefficient and the reaction rate
constants were analysed in order to provide a further understanding
of the mechanisms that control the reactive extraction process. Fig. 9
(a, b and c) shows the effect of independently increasing the reaction
rate constants, the diffusion coefficient and the external mass transfer
Monoglyceride concentration (mol/l)
coefficient respectively. The higher reaction rates of Komers et al.
(2002) [20] and Vicente et al. (2005) [14] were also used as a comparison.
It can be seen from Fig. 9a and c that higher reaction rate constants
and mass transfer coefficients did not significantly affect the overall ex-
traction rate, but increasing the diffusion coefficient (Fig. 9b) causes the
extraction rate to increase. This implies that the overall extraction rate
in these conditions is limited by diffusion rate rather than reaction
rate or external mass transfer. Hence, increasing the mixing speed, for
example, will not increase the overall extraction rate, as external mass
transfer resistance is insignificant. Likewise, increasing the rate of reac-
tion (i.e. by means of increasing the catalyst concentration greater than
0.1 m) is not expected to increase the rate of overall extraction. Given
this dependence on internal diffusion, the overall reactive extraction
rate would be expected to increase if the seed were treated in such a
way that the internal cell walls were damaged, or if the diffusional
path is shortened by reducing particle size.
Conversely, the model predicts that lowering the reaction rate con-
stants (i.e. by using lower catalyst concentration) decreases the overall
extraction rate as can be seen in Fig. 10. This agrees with actual experi-
mental results which show that decreasing the catalyst concentration
from 0.1 m to 0.05 m and 0.03 m significantly decreases the overall
rate of extraction. Hence, it can be deduced that the maximum catalyst
concentration is 0.1 m, above which the process rate is not affected
by catalyst concentration but is controlled by diffusional limitation.
Moreover, if it is assumed that the individual apparent rate con-
stants, k′1 to k′− 3, decrease by the same proportion when the catalyst
concentration is decreased, the model reveals that the rate constants
determined for 0.05 m catalyst concentration (in the bulk solvent)
was 50 times lower than at 0.1 m while the rate constant of 0.03 m
was 2000 times lower than at 0.1 m. Clearly the reaction rate decreases
exponentially with catalyst concentration and the dependent can be
represented by equation [16] below:

6.E-03 0 6:2746
k ¼ kC ð16Þ
5.E-03
1
4.E-03
0.5
3.E-03
0
2.E-03 0.00295 0.003 0.00305 0.0031 0.00315 0.0032 0.00325 0.0033 0.00335
log k

-0.5
1.E-03
-1
0.E+00
0 10 20 30 40 50
-1.5
Time (min)
-2
30 deg C - exp data 40 deg C - exp data 60 deg C - exp data 1/Temperature ( K-1)
30 deg C - model 40 deg C - model 60 deg C - model
k1 k2 k3 k4 k5 k6
Fig. 6. Effect of temperature on monoglyceride concentration profile. Reaction conditions:
25 g seed, 0.1 m catalyst concentration, 250 ml solvent. Fig. 8. Reaction rate constants at different temperatures.
 R. Zakaria, A.P. Harvey / Fuel Processing Technology 125 (2014) 34–40 39
Ester concentration (mol/l)
0.16
0.14 a
0.12
0.1
0.08
0.06
0.04
0.02
0
0 10 20 30 40 50 60 70
Time (min)
Experimental data Komers rate Vicente rate k 30k
Fig. 11. Effect of increasing diffusivity at low reaction rate (0.03 m catalyst concentration).
Ester concentration (mol/l)

0.16 This is probably because the concentration of catalyst inside the seed
0.14 b may not be the same as in the bulk solvent as some catalyst may be
0.12 absorbed by/reacted with other substances in the seed. Hence, lowering
0.1
the catalyst concentration in the bulk solvent may cause a much lower
0.08
catalyst concentration inside the seed, which significantly reduces the
0.06
0.04 overall extraction rate. Alternatively, the reaction rate inside the seed
0.02 may have a different dependency on the catalyst concentration due to
0 it being a surface reaction between the oil globule and methanol rather
0 20 40 60 80 100 120 140 than a homogenous reaction in the solvent. A study on reactive extrac-
Time (min) tion of municipal sludge using acid catalyts has also shown that reduc-
Exp data D=1.5e-11 D=3.5e-12 D=1.5e-12 ing the catalyst concentration from 5% to 1% significantly reduces the
rate of the reactive extraction process [21].
If the reaction rate constants are much lower than those of Fig. 8 (i.e.
Ester concentration (mol/l)

0.16 for 0.03 m catalyst concentration), the model shows that increasing the
0.14 c diffusion coefficient scarcely increased the rate of the extraction as
0.12 shown in Fig. 11. This is realistic, since at lower catalyst concentra-
0.1
tions, the rate of reactive extraction becomes predominantly reaction-
0.08
0.06
controlled. As a consequence, in this region, the rate of overall extraction
0.04 rate will not be greatly affected by efforts to increase the diffusivity in the
0.02 seed such as by physical or chemical seed pre-treatment.
0
0 20 40 60 80 100 120 140
Time (min) 5. Conclusions

Exp data Km=1.4e Km=1.4e-3
Fig. 9. (a) Effect of increasing reaction rates (k's are the reaction rate constants from Fig. 8).
(b) effect of increasing diffusivity and (c) effect of increasing external mass transfer
coefficent. Reaction conditions: 60 °C, 25 g seed, 0.1 m catalyst concentration, 250 ml
solvent.
where k is the effective reaction rate constant, k′ is the apparent reac-
tion rate constant and C is the catalyst concentration in mol/kg-solvent.
In comparison the apparent rate constant in conventional
transesterification decreases linearly with molar catalyst concentration,
as would be expected [14]. It seems that there is a much stronger de-
pendency on catalyst concentration in the reactive extraction process
if the reaction occurs inside the seed rather than in the bulk methanol.
Km=6.4e-4 The batch reaction time profile of the reactive extraction of biodiesel
from rapeseed particles was modelled by postulating that the reaction
occurs inside the seed and the products diffuse to the bulk solvent.
The model indicates that at high catalyst concentration, the overall reac-
tive extraction rate is controlled by the internal diffusion rate. This
means that the rate of the process can be improved if the diffusion coef-
ficient in the seed can be increased, i.e. by means of seed pre-treatment.
At low catalyst concentrations (below 0.1 mol/kg) the rate of the
process is controlled by the reaction rate. This agrees well with experi-
mental data, which shows that the overall extraction rate decreased
when the catalyst concentration was reduced from 0.1 to 0.3 mol/kg,
but is unaffected by increasing the concentrations above 0.1 m.
Increasing the external mass transfer coefficient (km) in the model to
above 1.4 × 10−4 m/s did not influence the rate. Hence, increasing the
degree of mixing would not increase the observed rate of this process.
The overall mass transfer coefficient increases with increasing tempera-
ture and the temperature dependence of the overall mass transfer coef-
ficient can be described by an Arrhenius relationship.

Acknowledgement

This work has been funded by the Ministry of Higher Education,

Malaysia and University Putra Malaysia.

References
[1] M. Canakci, J.H. Van Gerpen, Comparison of engine performance and emissions for
Fig. 10. Effect of lowering catalyst concentration on the ester concentration in bulk solvent. petroleum diesel fuel, yellow grease biodiesel, and soybean oil biodiesel, Transac-
Reaction conditions: 60 °C, 25 g seed, 250 ml solvent. tions of the American Society of Agricultural Engineers 46 (2003) 937–944.
40 R. Zakaria, A.P. Harvey / Fuel Processing Technology 125 (2014) 34–40
[2] C. Teas, S. Kalligeros, F. Zannikos, An investigation of using biodiesel/marine diesel
blends on the performance of a stationary diesel engine, Biomass and Bioenergy
24 (2002) 141–149.
[3] M. Balat, Fuel characteristics and the use of biodiesel as a transportation fuel, Energy
Sources, Part A: Recovery, Utilization, and Environmental Effects 28 (2006) 855–864.
[4] K.J. Harrington, C. D'Arcy-Evans, A comparison of conventional and in situ methods
of transesterification of seed oil from a series of sunflower cultivars, Journal of the
American Oil Chemists' Society 62 (1985) 1009–1013.
[5] J. Zeng, X. Wang, B. Zhao, J. Sun, Y. Wang, Rapid in situ transesterification of sunflow-
er oil, Industrial and Engineering Chemistry Research 48 (2009) 850–856.
[6] S. Özgül, S. Türkay, In situ esterification of rice bran oil with methanol and ethanol,
Journal of the American Oil Chemists' Society 70 (1993) 145–147.
[7] F.H. Kasim, A. Harvey, R. Zakaria, Biodiesel production by in situ transesterification,
Biofuels 1 (2010) 355–365.
[8] G. Kildiran, S.Ö. Yücel, S. Türkay, In-situ alcoholysis of soybean oil, Journal of the
American Oil Chemists' Society 73 (1996) 225–232.
[9] M.J. Haas, K.M. Scott, W.N. Marmer, T.A. Foglia, In situ alkaline transesterification: an
effective method for the production of fatty acid esters from vegetable oils, Journal
of the American Oil Chemists' Society 81 (2004) 83–89.
[10] J. Qian, F. Wang, S. Liu, Z. Yun, In situ alkaline transesterification of cottonseed oil for
production of biodiesel and nontoxic cottonseed meal, Bioresource Technology 99
(2008) 9009–9012.
[11] Y. Ren, A. Harvey, R. Zakaria, Biorefining based on biodiesel production: chemical
and physical characterisation of reactively extracted rapeseed, Biobased Materials
and Bioenergy 4 (2010) 1–8.
[12] George C. So, Douglas G. MacDonald, Kinetics of oil extraction from canola (rapeseed),
Canadian Journal of Chemical Engineering 64 (1986) 80–86.
[13] H. Noureddini, D. Zhu, Kinetics of transesterification of soybean oil, Journal of the
American Oil Chemists' Society 74 (1997) 1457–1463.
[14] G. Vicente, M. Martinez, J. Aracil, A. Esteban, Kinetics of sunflower oil methanolysis,
Industrial and Engineering Chemistry Research 44 (2005) 5447–5454.
[15] S.K. Karmee, D. Chandna, R. Ravi, A. Chadha, Kinetics of base-catalyzed
transesterification of triglycerides from pongamia oil, Journal of the American Oil
Chemists' Society 83 (2006) 873–877.
[16] R. Zakaria, A. Harvey, Direct production of biodiesel from rapeseed by reactive
extraction/in situ transesterification, Fuel Processing Technology 102 (2012)
53–60.
[17] U. Rashid, F. Anwar, Production of biodiesel through optimized alkaline-catalyzed
transesterification of rapeseed oil, Fuel 87 (2008) 265–273.
[18] D.A. Sasmaz, Evaluation of the diffusion coefficient of rapeseed oil during solvent
extraction with hexane, Journal of the American Oil Chemists' Society 73 (1996)
669–671.
[19] K.A. Reddy, L.K. Doraiswamy, Estimating liquid diffusivity, Industrial and Engineer-
ing Chemistry Fundamentals 6 (1967) 77–79.
[20] K. Komers, F. Skopal, R. Stloukal, J. Machek, Kinetics and mechanism of the KOH —
catalyzed methanolysis of rapeseed oil for biodiesel production, European Journal
of Lipid Science and Technology 104 (2002) 728–737.
[21] A. Mondala, K. Liang, H. Toghiani, R. Hernandez, T. French, Biodiesel production by in
situ transesterification of municipal primary and secondary sludges, Bioresource
Technology 100 (2009) 1203–1210