You are on page 1of 13

Applied Catalysis B: Environmental 67 (2006) 136148

www.elsevier.com/locate/apcatb

Simulation of heterogeneously MgO-catalyzed transesterification


for fine-chemical and biodiesel industrial production
Tanguy F. Dossin, Marie-Francoise Reyniers *, Rob J. Berger, Guy B. Marin
Laboratorium Voor Petrochemische Techniek, UGent Krijgslaan 281 (S5), B-9000 Gent, Belgium
Received 19 January 2006; received in revised form 5 April 2006; accepted 8 April 2006
Available online 5 June 2006

Abstract
A heterogeneous magnesium oxide catalyst is a good alternative for homogeneous catalysts for the transesterification of alkyl esters for the
production of fine-chemicals as well as for the production of biodiesel. The transesterification of ethyl acetate with methanol was used as a model
reaction to simulate fine-chemical production in a batch slurry reactor at industrial conditions. The transesterification of triolein with methanol to
methyl oleate was chosen to simulate continuous production of biodiesel from rapeseed oil. A kinetic model based on a three-step EleyRideal
type of mechanism in the liquid phase was used in both process simulations. The transesterification reaction occurs between methanol adsorbed on
a magnesium oxide free basic site and ethyl acetate or the glyceride from the liquid phase. Methanol adsorption is assumed to be rate-determining
in both processes. Activity coefficients were required to account for the significant non-ideality of the reaction mixture in the simulations of both
processes. The simulations indicate that a production of 500 tonnes methyl acetate per year can be reached at ambient temperature in a batch
reactor of 10 m3 containing 5 kg of MgO catalyst, and that a continuous production of 100,000 tonnes of biodiesel per year can be achieved at
323 K in a continuous stirred reactor of 25 m3 containing 5700 kg of MgO catalyst. Although various assumptions and simplifications were made
in these explorative simulations the assumptions concerning the reaction kinetics used, the results indicate that for both processes a heterogeneous
magnesium oxide catalyst shows promising potential as a viable industrial scale alternative.
# 2006 Elsevier B.V. All rights reserved.
Keywords: Transesterification; Solid base catalyst; Magnesium oxide; Reactor modeling; Industrial simulation; Biodiesel; Rapeseed oil; Kinetics

1. Introduction
Transesterification of alkyl esters plays an important
industrial role with numerous applications for large and small
volume productions. Large-scale applications are, e.g. the
production of biodiesel [1], polyester or PET in the polymer
industry [2,3] while small-scale fine-chemical production
includes synthesis of intermediates for the pharmaceutical
industry, the production of food additives or surfactants and the
curing of resins in the paint industry [2,4,5]. Nowadays, most
industrial applications are performed in batch or continuous

Abbreviations: BuOAc, butyl acetate; BuOH, n-butanol; BuOK, potassium


n-butanolate; D, diolein; EtOAc, ethyl acetate; EtOH, ethanol; FAME, fatty
acid methyl ester; G, glycerine; M, monoolein; M/E, methanol to ethyl acetate
molar ratio; M/T, methanol to triolein molar ratio; MeOAc, methyl acetate;
MeOH, methanol; MgO, magnesium oxide; MeOl, methyl oleate; T, triolein
* Corresponding author. Tel.: +32 9 264 45 17; fax: +32 9 264 49 99.
E-mail address: mariefrancoise.reyniers@ugent.be (M.-F. Reyniers).
0926-3373/$ see front matter # 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcatb.2006.04.008

reactors depending on the production volume. Temperatures


range from 333 to 523 K while pressure varies from 0.1 to
10 MPa and alcohol to ester molar ratios from 5 to 15.
Transesterification reactors are usually coupled with distillation
columns to separate the alcohols from the reaction mixture
[611]. Reactive distillation is also currently investigated as
transesterification process [1215]. Transesterification reactions are mainly performed using acid or base homogeneous
catalysts: sulfuric, sulfonic, phosphoric and hydrochloric acids
as acid catalysts [2,14] and alkaline hydroxides, metal
alkoxides or acetates as base catalysts [2,6]. Base catalysts
are preferred over acid catalysts because of the higher reaction
rates and the lower process temperatures [16]. However, the use
of homogeneous catalysts requires neutralization and separation from the reaction mixture, leading to a series of
environmental problems related to the use of large amounts
of solvents and energy. Heterogeneous solid base catalysts, able
to catalyze the transesterification of alkyl esters could solve
these issues: they can be easily separated from the reaction

T.F. Dossin et al. / Applied Catalysis B: Environmental 67 (2006) 136148

Nomenclature
ai
aLS
amn
A
Ci
d1,2
dI
dp
Di
Di,eff
Dim
D0AB
Ea
F
g
kBuOK
kMeOH
kLS,i
Keq
KA
li
MWi
ni
N
NI
NP
p
P
q
qi
rBuOK
ri
rMgO
R
Re
Sci
t
T
VL
W
xi
Xi
Yi

activity of component i (mol m3)


liquidsolid interfacial area (m2 m3)
group interaction parameter (in calculation
activity coefficient)
pre-exponential factor (m3 kgcat1 s1)
concentration of component i (mol m3)
FloryHuggins combinatorial term
impeller diameter (m)
solid particle diameter (m)
liquid diffusion coefficient of component i
(m2 s1)
effective diffusion coefficient of component i in
the pellet pores (m2 s1)
bulk diffusivity of component i (m2 s1)
binary diffusivity of component B in A (m2 s1)
activation energy (J mol1)
molar flow (mol s1)
acceleration of the gravity (m2 s1)
reaction rate coefficient for BuOK catalyzed
reaction (m6/mol2 s)
reaction rate coefficient of methanol adsorption
step (m3 kgcat1 s1)
liquidcatalyst mass transfer coefficient for component I (m3 m2 s1)
equilibrium constant of the overall reaction
equilibrium constant of alcohol adsorption
(m3 mol1)
volume parameter (in calculation activity coefficient)
molecular weight of component i (kg/mol)
number of moles of component i (mol)
total number of components
impeller revolution speed (s1)
impeller power number (=5) (s1)
pressure (atm)
yearly production (tonnes year1)
induction parameter
molecular surface area parameter (in calculation
activity coefficient)
BuOK-catalyzed reaction rate (mol m3 s1)
van der Waals volume parameter (in calculation
activity coefficient)
MgO-catalyzed reaction rate (mol kgcat1 s1)
universal gas constant (J mol1 K1)
Kolmogoroff Reynolds number = r3L d 4p N p dI5 Ni3 =
m3L VL
Schmidt number component i = kLS,i dp/Dim (s)
time (s)
temperature (K)
liquid volume (m3)
mass of catalyst (kgcat)
molar fraction of component i
conversion of component i
product yield of component i

137

Greek symbols
a
acidity parameter
b
basicity parameter
gi
activity coefficient of component i
g1
activity coefficient of component i at infinite
i
dilution
g Ci
combinatorial contribution to activity coefficient
of component i
g Ri
residual contribution to activity coefficient of
component i
Fi
segment fraction (in calculation activity coefficient)
ui
area fraction (in calculation activity coefficient)
Gk
group residual activity coefficient
i
Gk
group residual activity coefficient in a reference
solution containing only molecules of type i
ep
catalyst porosity
es
solid fraction in the slurry reactor
l
dispersion parameter
Li,j
Wilson binary parameter of component i in
solvent j
mL
liquid viscosity (kg m1 s1)
n
liquid molar volume at 293 K (cm3 mol1)
j
asymmetry parameter due to hydrogen bonding
rL
liquid density (kg m3)
rp
catalyst pellet density (kg m3)
rp,wet density of the catalyst particle filled with liquid
(kg m3)
t
polar parameter or catalyst tortuosity
c
asymmetry parameter due to polarity difference
cmn
group interaction parameter (in calculation
activity coefficient)
Subscripts and superscripts
0
initial condition
*
basic site
cat
catalyst
eq
equilibrium
i
component i
in
inlet
I
impeller
j
parameter j
k
kth experimental point
out
outlet
p
catalyst pellet
R
reactor
s
solvent
tot
total

mixture without requiring the use of a solvent, show easy


regeneration, and have a less corrosive character, leading to
safer, cheaper and more environment-friendly operation.
Recently, numerous studies have been performed to investigate
the use of heterogeneous base compounds as catalysts for
transesterification reactions. This includes the use of catalysts

138

T.F. Dossin et al. / Applied Catalysis B: Environmental 67 (2006) 136148

such as alkaline-earth oxides, zeolites, and hydrotalcites


[3,4,1720]. Nowadays only one industrial application,
developed by the IFP, uses heterogeneous base catalysts for
the transesterification of vegetable oils to produce biodiesel
[21].
Although several authors investigated the kinetics of
transesterification catalyzed by homogeneous base catalysts
[6,14,16,2224], there is very little information available
concerning the kinetics of the transesterification catalyzed by
solid bases. Hattori et al. [18] have proposed a mechanism for
the transesterification of ethyl acetate with different alcohols
catalyzed by a variety of solid base catalysts, particularly
alkaline-earth-metal oxides. However, this mechanistic study
did neither provide values of activation energies nor rate
constants. Recently, a kinetic model was proposed to describe
the transesterification of ethyl acetate with methanol catalyzed
by a heterogeneous magnesium oxide catalyst [25]. This model
is based on a three-step EleyRideal type of mechanism
applied in liquid phase where methanol adsorption is assumed
to be rate determining. The proposed value of the activation
energy is 20 kJ/mol and the value of the methanol adsorption
equilibrium coefficient is 3.13  103 m3/mol. From several
kinetic studies on homogeneous base catalysts followed a
pseudo-second-order rate law with activation energies ranging
from 22 to 83 kJ/mol, depending on the type of alcohol and
ester used [6,14,16,22,23].
The objective of this study was to evaluate the use of the
heterogeneously MgO-catalyzed transesterification reaction in
batch and continuous stirred tank reactors at industrially
relevant conditions using the kinetic model based on the threestep EleyRideal type mechanism assuming methanol
adsorption as rate-determining step. Two processes have been
simulated: the transesterification of ethyl acetate with methanol
in a batch slurry reactor for fine-chemical production and the
transesterification of triolein with methanol to form methyl
oleate to simulate biodiesel production from rapeseed oil in a
continuous slurry reactor. The simulation results from both
have been compared with those of homogeneously catalyzed
transesterification processes.
Biodiesel consists of alkyl esters of long chain fatty acids
originating from a renewable resource such as vegetable oils.
Biodiesel can be used in usual diesel engines and presents some
advantages compared to traditional petroleum-based diesel.
Biodiesel is an environment-friendly compound, non-toxic and
emits less carbon monoxide, sulphur compounds, particulate
matter and unburned hydrocarbons than traditional diesel.
However, NOx emissions are 10% higher compared to
petroleum-based diesel. Moreover, since biodiesel is produced
from a renewable source, the net production of carbon dioxide
is lowered with reduced contribution to the greenhouse effect.
Biodiesel is usually produced by transesterification of
vegetable oils, largely consisting of triglycerides, with
methanol to produce fatty acid methyl esters (FAME), i.e.
biodiesel, and glycerine.
The catalysts used are typically alkali hydroxides such as
KOH or NaOH and alkali alkoxides such as sodium methoxide
or sodium butoxide. Non-ionic base catalysts such as TBD

(1,5,7-triazabicyclo[4.4.0]dec-5-ene, C7H13N3) are currently


studied [20,26]. The main purpose of transesterification is to
reduce the viscosity of the oil. Vegetable oils cannot be used as
such in common diesel engines because their viscosity is 1020
times that of conventional diesel, causing injector coking and
engine deposits [1,26]. Biodiesel does not show these problems
and offers the same performance and engine durability as
petroleum-based diesel. Therefore, the importance of biodiesel
increases due to environmental concerns and uncertainties
concerning petroleum price and availabilities.
The transesterification of triolein with methanol to methyl
oleate (CH3(CH2)7CH=CH(CH2)7COOCH3) has been chosen
as a model process for the transesterification of rapeseed oil to
produce biodiesel. Rapeseed oil is the most abundant oil
produced in Europe and triolein (C57H104O6) has been chosen
to represent rapeseed oil as oleic acid is the major fatty acid in
rapeseed oil [26].
2. Modeling procedures
2.1. Transesterification of ethyl acetate with methanol

2.1.1. Thermodynamics
Thermodynamic properties of the components and the
transesterification reaction between 283 and 323 K were
calculated using the ASPEN Engineering SuiteTM 11.1 [27].
The reaction enthalpy varies from 2.8 at 273 K to 3.2 kJ/
mol at 328 K, showing a slightly exothermic reaction. The
value of the equilibrium coefficient Keq ranges from 1.875 at
293 K to 1.716 at 323 K, which is similar to values reported in
previous studies of transesterification reactions [14,22]. The
equilibrium conversion of ethyl acetate Xeq varies from 9 to
95% at molar ratios of methanol to ethyl acetate (M/E) ranging
from 0.1 to 10.
The non-ideal character of the reaction mixture was assessed
and activity coefficients gi for each component i were
calculated at temperatures ranging from 283 to 323 K based
on the Wilson equation as follows [28]:
ln g i ln

X
N
j1


x j Li; j 1 

N
X

xk Lk;i
PN
k1
j1 x j Lk; j

(1)

where, N, xi and Li,j are the number of components, the mole


fraction of each component and the Wilson binary parameter of
component i diluted in solvent j. The Wilson binary parameters
are calculated based on Eq. (1) assuming infinite dilution of
component i in solvent j. The activity coefficient at infinite
dilution is then calculated based on Eq. (2) by means of the
MOSCED method [28]:
ln g 1
i


ni
q2 q2 ts  ti 2

ls  li 2 s i
RT
cs

as  ai bs  bi

ds;i
js

(2)

T.F. Dossin et al. / Applied Catalysis B: Environmental 67 (2006) 136148


Table 1
Influence of the M/E ratio on the activity coefficients of MeOH, EtOH, MeOAc,
EtOAc at 303 K, at the start of the reaction and at equilibrium
M/E ratio

0.1

10

Conversion (mol%)

58

89

95

MeOH
EtOH
MeOAc
EtOAc

3.08

1.02

3.11
2.79
1.00
1.02

1.29

1.42

1.30
1.26
1.30
1.42

1.03

2.22

1.03
1.09
1.90
2.24

1.01

2.51

1.01
1.09
2.10
2.53

where g 1
i is the activity coefficient of component i in solvent s
at infinite dilution, and a, b, l, n, j, t, c, q and ds,i are tabulated
parameters, specific for each component.
Table 1 shows the influence of the M/E ratio at 303 K on the
activity coefficients of each component at conditions corresponding to the start of the reaction and to the equilibrium
conversion. Since the calculated activity coefficients were
fairly different from 1.0, the non-ideality of the mixture was
taken into account in the simulations.
2.1.2. Kinetic model
The kinetic model used to describe the transesterification of
ethyl acetate with methanol to form methyl acetate and ethanol
is based on a three-step EleyRideal type mechanism [25].
The elementary steps of the mechanism are given in Table 2.
The first step describes the adsorption of methanol on a
catalytically active site. Then, adsorbed methanol reacts with
ethyl acetate from the liquid phase to form methyl acetate and
adsorbed ethanol. Ethanol finally desorbs in the third step. The
proposed model assumes methanol adsorption as rate-determining step, leading to the following rate equation:
rMgO

kMeOH aMeOH  1=Keq aMeOAc aEtOH =aEtOAc


1 KA =Keq aMeOAc aEtOH =aEtOAc KA aEtOH

(3)

where kMeOH is the methanol adsorption rate constant, ai, the


activities of reaction component i, Keq the equilibrium coefficient of the overall transesterification reaction and KA is the
equilibrium constant of the methanol adsorption step. The value
of the adsorption equilibrium coefficient is assumed to be the
same for both alcohols. The values of the activation energy Ea
and the pre-exponential factor A of the rate constant kMeOH as
well as the adsorption equilibrium coefficient KA have been
estimated and their values are presented in Table 3. The
parameter estimation procedure and its results are described
elsewhere [25]. The value of Keq has been calculated based on
thermodynamic data for all compounds.
According to Madon and Iglesia [29], it is appropriate in
non-ideal systems to use the activities instead of the
concentrations in the rate expression if the rate-determining
Table 2
Elementary reactions of EleyRideal type reaction mechanism
CH3OH*
CH3OH + *
CH3OH* + CH3COOCH2CH3
CH3CH2OH* + CH3COOCH3
*
*
CH3CH2OH
CH3CH2OH +

139

Table 3
Parameter values of the kinetic model used
A (m3/kgcat s)
Ea (103 J/mol)
KA (103 m3/mol)

0.148
20.1
5.29

step is an adsorption or desorption step, because the


corresponding activated complexes may be solvated by the
surrounding fluid phase. However, if the solvation effect of the
activated complex is similar to that of the reactants (products),
the concentrations instead of the activities should be applied
because a cancellation of activity coefficients leads to the
apparent dependence of rates on concentration rather than
activities. For the current reaction, it is unclear which situation
applies and therefore the effect of this was investigated, thus
using the following rate equation:
rMgO

kMeOH CMeOH  1=Keq aMeOAc aEtOH =aEtOAc


1 KA =Keq aMeOAc aEtOH =aEtOAc KA CEtOH

(3a)

Since the location of the thermodynamic equilibrium in nonideal mixtures is determined by the equilibrium constants in
combination with the component activities rather than the
component concentrations, the activities are maintained in
the terms correcting for the equilibrium. This thus results in
a rather unusual rate expression containing activities as well as
concentration terms. A similar form of rate equation, however,
has been proposed by Madon and Iglesia [29] for the cyclohexene hydrogenation over Pd.
Simulations results using rate Eq. (3a) are compared to
simulation results using Eq. (3) to study the possible influence
of solvation in the non-ideal liquid phase.
2.1.3. Reactor model
The slurry reactor, operated batch wise or in a continuous
way, was chosen as industrial reactor because of its easy
operability and frequent use in fine-chemical industry. The

Table 4
Range of process parameters
Conditions
T (K)
p (atm)
M/E ()
W (kg)
in
(mol s1)
FEtOAc

293323
1
0.130
510000
1100

Reactor
VL (m3)
dR (m)
hR (m)
NI (s1)
dI (m)

1040
2.33.7
2.33.7
1
1.21.9

Catalyst
dp (106 m)
ep
tp
rp (kg/m3)

25
0.45
5.0
2460

140

T.F. Dossin et al. / Applied Catalysis B: Environmental 67 (2006) 136148

reactor was considered to be cylindrical with equal reactor


diameter and height and with a volume between 10 and 40 m3.
The range of process parameters considered in the simulations
is given in Table 4. Since transesterification reactions are
equilibrium-controlled reactions, an excess of methanol is used
to shift the equilibrium towards the product side. The methanol
to ethyl acetate molar ratio (M/E) varies from 5 to 30 [16]. The
reactor geometry follows the standard tank geometry for
turbulent mixing with a radial flow impeller proposed by
Tatterson [30]. The impeller is a traditional Rushton turbine
with a diameter from 1.2 to 1.9 m. The impeller speed is
calculated to fulfill perfect mixing and complete suspension of
solid particles, following Eq. (4) [31]:


g
NI > 12
dI

0:5 

r p;wet  rL
rL

0:5 

dp
dI

0:165

e0:25
S

(4)

where NI is the impeller speed, dI, the impeller diameter, g, the


gravitational constant, r p;wet , the density of the catalyst pellet
filled with liquid, rL, the liquid density, dp, the particle diameter
and es is the catalyst fraction in the slurry. Since the criterion
was only validated in the range 0.01 < eS < 0.3, a value of
eS = 0.01 instead of the actual value of 0.00081 (0.5 kg catalyst/
m3) was used to calculate the impellor rate, and the resulting
minimum required impeller rate was interpreted as a conservative estimate. Based on this equation and using a particle
diameter of 25 mm, the minimum required impeller speed
varies between 0.45 and 0.57 s1, depending on the reactor
volume. This is in agreement with typical industrial impeller
speeds. For the simulations, an impeller speed of 1.0 s1 was
chosen.
As the reaction is only slightly exothermic, isothermal
behavior was assumed. The influence of temperature was
studied from 293 K (ambient temperature) to 323 K since the
boiling point of methyl acetate is 329 K. The simulations were
performed at atmospheric pressure.
From the reactor geometry, the kinetic data, and the physical
properties of the catalyst and the reaction components, the
inter- and intra-particle mass and heat transport limitations
were estimated for the transesterification of ethyl acetate with
methanol catalyzed by MgO, based on correlations available in
the literature [3234]. Intra-particle mass and heat transport
limitations appeared to be negligible with a pellet diameter not
larger than 25 mm. This also leads to a uniform temperature
inside the pellet.
Several assumptions were adopted for the simulations.
First, since isothermal behavior was assumed and since heat
transfer limitations are negligible, the temperature was
assumed to be uniform inside the reactor. Secondly, the molar
volume and also the total volume of the reaction mixture was
assumed constant since the reaction takes place only in the
liquid phase and the total number of moles remains constant.
Additionally, no catalyst deactivation was considered since
catalyst deactivation only occurs if the catalyst is contacted
with water, CO2 or air [35,36], which can be avoided if the
reaction occurs under a protected atmosphere like nitrogen or
argon atmosphere. A detailed investigation of catalyst leaching

and solubility has been performed to assess the heterogeneous


character of the catalyzed reaction and is described elsewhere
[25].
Finally, batch time is assumed to be the reaction time; the
time required for filling and emptying the reactor is not
considered.
Since the experimental study did not reveal the production of
any byproducts [25], their formation was not considered in the
simulations. The batch slurry reactor was therefore simulated
by considering the continuity equation of one reaction
component only, e.g. ethyl acetate:
dnEtOAc
rW
dt

(5)

where nEtOAc is the number of mole of ethyl acetate, W the mass


of the catalyst and r is the specific reaction rate (mol/kgcat s).
The other components in the reaction mixture are expressed as a
function of nEtOAc. The integration of the ordinary differential
Eq. (5) was performed using the LSODA-subroutine [37]
available at Netlib [38]. A typical production volume of
fine-chemicals in a batch slurry reactor is 500 tonnes
MeOAc year1. As a criterion for equilibrium, a relative deviation of 5% from the ethyl acetate conversion at equilibrium was
used.
2.2. Transesterification of triolein with methanol
The transesterification of triolein by methanol consists of the
following three consecutive transesterification reactions [16]:
T MeOH D MeOl

(6)

D MeOH M MeOl

(7)

M MeOH G MeOl

(8)

Firstly the triglyceride, triolein (T), reacts with the alcohol, in


this case methanol (MeOH), to produce a diglyceride, diolein
(D), and the fatty acid ester, in this case methyl oleate (MeOl).
Then the diglyceride formed in the first reaction reacts further
with the alcohol to form another molecule of the fatty acid ester
and a monoglyceride, monoolein (M). This then reacts with the
alcohol to obtain glycerine (G) and the fatty acid ester. This
leads to the overall reaction:
T 3MeOH G 3MeOl

(9)

where the triglyceride molecule reacts with three alcohol


molecules to produce three molecules of methyl ester and
glycerine.
2.2.1. Thermodynamics
Since glycerides are complex molecules for which no
thermodynamic properties could be found, the thermodynamic
properties were estimated using the Benson contribution
method [28]. The calculated equilibrium constants Keq,1,
Keq,2, and Keq,3, for the glyceride transesterification reactions of
Eqs. (6)(8) appeared to vary between 0.98 and 1.02 in the
temperature range 313343 K. It is not surprising that these

T.F. Dossin et al. / Applied Catalysis B: Environmental 67 (2006) 136148

141

Table 5
Influence of the M/T ratio on the activity coefficients (gi) of all components at 323 K, at the start of the reaction and at equilibrium (xi = mole fraction)
Triolein

Diolein

Monoolein

Methyl oleate

Glycerine

MeOH

M/T = 5
xi
gI
xi (equilibrium)
gI

0.16648
1.34479
0.01987
1.23714

0.00004
0.65299
0.06435
0.60129

0.00006
0.83860
0.07885
0.77292

0.00029
2.15770
0.23293
2.06543

0.00003
28.27353
0.00368
26.08368

0.83311
1.23042
0.60033
1.26260

M/T = 10
xi
gI
xi (equilibrium)
gI

0.09086
2.58711
0.00432
2.35993

0.00000
0.71287
0.02538
0.64583

0.00002
0.51951
0.05645
0.46744

0.00012
3.67922
0.15259
3.50896

0.00003
9.93927
0.00478
8.88203

0.90897
1.12314
0.75648
1.13984

M/T = 20
xi
gI
xi (equilibrium)
gI

0.04759
6.56754
0.00077
6.22731

0.00000
0.97234
0.00826
0.90549

0.00001
0.38074
0.03344
0.34822

0.00007
6.66597
0.09059
6.45176

0.00002
3.91396
0.00516
3.51569

0.95231
1.05075
0.86177
1.05628

values are close to 1.0 since the reactions are almost


thermoneutral and the entropy change is likely to be very
small as well. For simplicity, all the equilibrium constants were
taken equal to 1.0 in the simulations, independent of
temperature.
Since it was observed that the reaction mixture consisting of
ethyl acetate, methanol, methyl acetate and ethanol was not
ideal (see Section 2.1.1), the need for taking non-ideality into
account was also checked for the biodiesel simulations.
However, unlike the transesterification of ethyl acetate with
methanol, the MOSCED method could not be applied to solve
the Wilson equation since data for glyceride molecules are not
tabulated. Therefore, the UNIFAC contribution method was
used to calculate activity coefficients [28]. The activity
coefficient is separated into two parts: one part provides the
contribution due to molecular size and shape and is defined as
the combinatorial contribution (ln g Ci ), and the other provides
the contribution due to molecular interactions, known as the
residual contribution (ln g Ri ):
ln g i ln g Ci ln g Ri

(10)

where
ln g Ci ln
ln g R

all comp:
Fi
ui
Fi X
5qi ln li 
x jl j
xi
Fi
xi
j
allX
groups

and

ni ln G k  ln G k ;

x i qi
ui P
;
j x jq j

x i ri
Fi P
;
j x jr j

li 5ri  qi  ri  1;
qi

X
k

ri

X
k

i
nk Qk

n k Rk ;

and

allX

comp:
ln G k Qk 1  ln
um C mk
m

allX
comp:
m

C mn

um C km

Pall comp:
n


amn
exp 
:
T

un C nm


;

Rk and Qk are group volumes and area parameters and are


tabulated as a function of the type of group. T is the temperature
i
expressed in Kelvin, nk is the number of groups k present in the
molecule i and xi is either the molar fraction of component i in the
mixture or the mole fraction of group i in the component. Table 5
shows the influence of the M/T ratio at 323 K on the activity
coefficients of each component and at the start of the reaction and
at equilibrium in terms of mole fraction. Since the calculated
activity coefficients were fairly different from 1.0, the nonideality of the mixture was taken into account in the simulations.
2.2.2. Kinetic model
Several kinetic studies of methyl ester production from
different vegetable oils have been performed using homogeneous base catalysts [6,16,20,23]; however, to the best of our
knowledge, no kinetic data of vegetable oil transesterification
involving heterogeneous catalysts have been reported. All
kinetic studies performed with homogenous catalysts show that
all three consecutive transesterification reaction rates follow a
pseudo-second-order law with respect to the glyceride and the
alcohol. As can be seen in Freedmans work [16], the three
consecutive transesterification reactions have significantly
different activation energies.
Since the reaction mechanism of the transesterification
reactions over the MgO catalyst is assumed to be of the
EleyRideal type with the adsorption of methanol being the

142

T.F. Dossin et al. / Applied Catalysis B: Environmental 67 (2006) 136148

rate-determining step, the kinetic model used for simulation of


the three heterogeneous transesterification reactions over the
MgO catalyst is assumed to be the same as that for the
transesterification of ethylacetate, as discussed in Section 2.1.2.
As this mechanism is of the EleyRideal type with the
adsorption of methanol being the rate-determining step, it can
be assumed that the values of the activation energy Ea and the
pre-exponential factor A for each of the three transesterifications are equal to the values for the transesterification of ethyl
acetate with methanol, as presented in Table 3. The rate
equations for the three glyceride transesterifications are thus
assumed to be, analogous to Eq. (3):


aD
kMeOH aMeOH  K1eq aMeOl
aT
rT
(11a)
aD
KA aD KA aM KA aG
1 KKeqA aMeOl
aT

kMeOH aMeOH  K1eq
rD

1 KKeqA

aMeOl aM
aD

1 KKeqA

aMeOl aG
aM

aMeOl aG
aM

(11c)

where kMeOH is the methanol adsorption rate constant equal for


all three reactions, ai, the activities of each reaction component,
Keq the equilibrium coefficient of the overall transesterification
reaction and KA is the adsorption equilibrium constant of the
alcohols present, assumed to be the same for all alcohols
present. The elementary steps and the corresponding reaction
mechanism for each transesterification reaction are shown in
Table 6.
It should be mentioned that the assumption of exactly the
same reaction kinetics for the transesterification of large
glyceride molecules as that for the transesterification of ethyl
acetate is rather speculative. The long organic tails of the
glyceride molecules will probably have some influence on the
kinetics. Additionally, the assumption of equal rate coefficients
for the three consecutive reactions seems rather speculative in
view of the significant differences between the activation
energies of the three consecutive transesterification reactions
observed with homogeneous catalysts, as mentioned above.
Table 6
Elementary steps and reaction mechanisms for the biodiesel transesterification
reactions
Reaction mechanism

CH3OH
CH3OH + *
CH3OH* + T
D* + MeOl
*
CH3OH + D
M* + MeOl
CH3OH* + M
G* + MeOl
*
D+*
D
M*
M+*
G*
G+*

rD

A
1 KKeq;2

aMeOl aM
aD

aMeOl aG
aM

KA CD KA CM KA CG


1
kMeOH CMeOH  Keq;3
A
1 KKeq;3

aMeOl aM
aD

aMeOl aG
aM

(12b)

KA CD KA CM KA CG

(12c)

(11b)

KA aD KA aM KA aG

Elementary reactions


1
kMeOH CMeOH  Keq;2

rM

KA aD KA aM KA aG


kMeOH aMeOH  K1eq
rM

aMeOl aM
aD

Similarly as for batch slurry reactor simulations (see Section


2.1.2), the results obtained using the rate equations based on
the component activities (11a)(11c) were compared to the
results using the rates based on the component concentrations
(Eqs. (12a)(12c)):


1 aMeOl aD
kMeOH CMeOH  Keq;1 aT
rT
(12a)
A aMeOl aD
KA CD KA CM KA CG
1 KKeq;1
aT

T!D

D!M

M!G

1
1

1
1
1
1
1

2.2.3. Reactor model


As for the simulation of fine-chemical production, a slurry
reactor has been chosen to simulate continuous biodiesel
production. All assumptions mentioned for the simulations in
the EtOAc/MeOH case are also valid for the biodiesel
production process (see Section 2.1.3). Additionally, the
continuous slurry reactor was assumed to be at steady state,
start-up time was not taken into account and the catalyst was
retained in the reactor by a microporous filter.
For simplicity reasons, the possibility of phase separation
between alcohols and glycerides was not included in the
explorative simulations discussed in this paper. According to
Noureddini and Zhou [23], this phase separation disappears
rapidly if a high impeller speed of at least 5 s1 is applied since
methyl oleate functions as mutual solvent to form a singlephase system as reaction advances. However, it seems more
likely that the good mixing leads to a micro-emulsion of a polar
and a non-polar liquid phase [1]. Particularly the high activity
coefficients of methyl oleate and glycerol (Table 5), which were
calculated assuming one homogenous liquid phase, indicate the
possibility of such a phase separation.
The existence of a phase separation may strongly influence
the reactor performance. On the one hand it may negatively
influence the performance if mass transport limitations occur
between both phases. On the other hand it may influence the
overall equilibrium composition, since a phase separation
generally leads to a significant decrease of the values of the
activity coefficients. Table 5 shows the composition at
equilibrium as a function of the M/T if no phase separation
is taken into account. For instance, at a M/T ratio of 10, the
maximum achievable mole fraction of methyl oleate amounts to
0.1526, corresponding to a yield of only 0.56. According to
these calculations, the reaction product will thus always contain
significant amounts of diolein and monoolein besides methyl
oleate, as shown in Table 5.
Experimental studies using homogeneous alkali catalysts,
however, have shown that methyl ester yields of over 0.9 can be

T.F. Dossin et al. / Applied Catalysis B: Environmental 67 (2006) 136148

reached at an M/Tabove 5 [23,24,46,39]. Besides the existence of


a phase separation, this discrepancy may also originate from the
estimation of the equilibrium constants or from the UNIFAC
contribution method used to estimate the activity coefficients.
Nevertheless, to the best of our knowledge, the UNIFAC
contribution method is currently the best method to estimate
activity coefficients of complex and non-tabulated molecules.
The occurrence of phase separation therefore seems to be the
most likely explanation of the discrepancy since the activity
coefficients of the final two products, methyl oleate and glycerol,
are the highest of all components, thus limiting the equilibrium
conversion to a relatively low value, which can be understood
based on Eq. (11c); a decrease of the activity coefficients as a
result of a phase separation will thus increase the conversion to
the desired product methyl oleate at equilibrium. Due to the
significant deviation of the equilibrium conversion, the simulation results are to be interpreted in a qualitative way only.
Another important aspect in this process is the presence of
mass transport limitations around and inside the catalyst
particles as a result of the relatively low diffusivity, which is
caused by the high viscosity of the reaction mixture. It may give
rise to significant concentrations gradients. In contrast, heat
transport limitations do not have to be taken into account as a
result of the thermoneutrality of the reaction. In order to take
the effect of mass transfer limitation into account, the reactor
model equation in the bulk phase for a continuous slurry reactor
considering steady-state operation and mass transport limitations is as follows:
Fiin  Fiout VL kLS;i aLS CL;i  CS;i
where Fiin

(13)

and Fiout

are inlet and outlet molar flows of compound


i, kLS,i the liquidsolid mass transfer coefficient of the component i in methanol, assumed to be in excess in the reactor, and
aLS the liquidcatalyst interface area, and VL is the total liquid
volume. CL,i and CS,i are the concentrations in the liquid phase
and at the external pellet surface, respectively.
The mass transfer coefficient is calculated from Sano et al.
[40]:

143

Since uniform spherical catalyst particles are assumed, the


mass balance inside the pellet is, applying spherical coordinates:


rp
4 1 @
@Ci
2
j
r
D
(15)
eff;i
2
2
e p d p j @j
@j
ep
Deff,i and Ci are the effective diffusion coefficient and the
concentration of component i in methanol inside the pellet;
ep and rp are the porosity and the density of the catalyst pellet,
respectively. j is the spatial coordinate inside de pellet, equal to
1 at the surface and 0 at the centre of the pellet. r is the net rate
of formation of this specific component. The effective diffusivity Deff,i depends on the concentrations through the dependence of the activity coefficients on the mole fractions. The
effective diffusion coefficient was estimated from [41]:
Deff;i

ep
Dim
tp

(16)

The bulk diffusivity Dim was calculated from the binary diffusion coefficient in methanol (A = i, B = methanol) [42] and
corrected for the effect of non-ideality [43] using:


@ ln g A
DAB xA D0BA xB D0AB 1
(17)
@ ln xA
Since xA D0BA  xB D0AB , this equation was simplified to:


@ ln g A
DAB xB D0AB 1
@ ln xA

(18)

resulting in:
Dim

xMeOH D0im



@ ln g i
1
@ ln xi

(19)

(14)

At the pellet surface, the effective diffusion of the component i


inside the pellet is equal to the diffusion of this component from
the bulk phase to the pellet surface. This is expressed by the
following equation:

2Deff;i @Ci 
CL;i  CS;i
(20)
kLS;i d p @j j1

The required liquid density and viscosity were calculated as the


weight-weighted average value of the values of the individual
components, which were taken from various sources as indicated
in Table 7.

Eq. (13) was applied to all reaction compounds and the reaction
rate r in this equation is the sum of the three transesterification
reaction rates (Eqs. (11a)(11c)) taking the reaction stoichiometry into account. Simulations were performed with the

kLS;i 2 0:4Re0:25 Sc0:333


i

Table 7
Pure component densities and viscosities used in the biodiesel production process simulations
Component

Densitya (kg/m3)

Source

Viscosityb (mPa s)

T0 (K)

Source

Triolein
Diolein
Monoolein
Methyl oleate
Glycerine
Methanol

11130.684T
10420.42T
11080.543T
10820.712T
14530.656T
10790.971T

[27,47]
[27]
[27]
[27,47]
[27,51]
[27,51]

45.1
68
91.5
5.2
1500
0.55

313
313
313
313
293
298

[47,48]
[49]
[49]
[48,47,50]
[51]
[51]

T: temperature (K).
0
The Lewis and Squires liquid viscositytemperature correlation [28] was used to obtain the viscosity at the required temperature: m0:2661
m0:2661
TT
T
T0
233 , with
mT = liquid viscosity at the required temperature T (K), and mT0 = known viscosity at temperature T0 (with viscosities expressed in mPa s).
b

144

T.F. Dossin et al. / Applied Catalysis B: Environmental 67 (2006) 136148

Athena Visual Workbench software developed by Stewart &


Associates Engineering Inc., using the PDAPlus solver. This
software allows the solution of boundary value problems
involving partial differential equations. The integration of
the catalyst particles was performed using the method of finite
differences with 150 discretization points.
The biodiesel production in a continuous slurry reactor
typically amounts to 100,000 tonnes year1 [44,45]. The
applicable temperature range is narrow due to the high
viscosity of the glycerides and the low boiling point of
methanol. Since the melting point of the monoglyceride is
about 310 K and the methanol boiling point is about 340 K, the
operating temperature window only ranges from 320 to 330 K.
Nowadays, the methanol to oil molar ratio is typically set to 6,
although a recent patent showed that higher molar ratios can
improve the yield sufficiently to make it economically
attractive [46]. Therefore, the operating molar methanol/
triolein ratio will be in the range between 5 and 30.
The conversions and the yields were calculated according to:
xT 1 
YMeOl
YG

FT;out
;
FT;feed

FMeOl;out
;
3FT;feed

xMeOH 1 
YD

FMeOH;out
FMeOH;feed

FD;out
;
FT;feed

YM

(21)
FM;out
;
FT;feed

FG;out
FT;feed
(22)

The yearly production of methyl oleate and glycerine were


calculated as follows:
PMeOl 3 

YMeOl  W  MWMeOl  60  60  24  365


1000W=FT;feed

tonnes=year
PG

(23)

YG  W  MWG  60  60  24  365
tonnes=year
1000W=FT;feed
(24)

3. Results and discussion


3.1. Batch slurry reactor for fine-chemical production
Simulations of a batch slurry reactor for the heterogeneously
MgO-catalyzed transesterification of ethyl acetate with
methanol at industrial process conditions have been performed
using Eq. (5). Influences of temperature, amount of catalyst and

molar ratio have been investigated to achieve a typical


production of 500 tonnes MeOAc year1. Traditionally, batch
times do not exceed 2 days.
Table 8 shows the evolution of the annual production as a
function of temperature ranging from 293 to 323 K, at M/E = 10,
W = 2 kg, VL = 10 m3. Fig. 1 shows the corresponding ethyl
acetate conversion as a function of the batch time. The minimum
required batch time is defined as the time to reach 95% of the
ethyl acetate equilibrium conversion. At the conditions stated
above, this corresponds to an EtOAc equilibrium conversion of
90%. As expected, the minimum required batch time decreases
with increasing temperature. For the considered process
parameters, a yearly production of almost 500 tonnes year1
is already achieved at 323 K (see Table 8). Nevertheless, it is
useful to study the influence of the other process parameters such
as the amount of catalyst and the M/E molar ratio on the yearly
production. The influence of the catalyst mass on the yearly
production is presented in Table 9. From these results follows that
a similar yearly production (524 tonnes year1) can be achieved
at 293 K if the amount of MgO catalyst is increased from 2 to
4.5 kg (M/E = 10, VL = 10 m3). Running the process at 293 K
could eliminate the need for reactor heating, which may make the
process easier and less expensive, assuming the cost of the
catalyst is much smaller than the cost of the process equipment
combined with the energy cost. The influence of the M/E ratio on
the minimum required batch time at 293 K, W = 5 kg,
VL = 10 m3 is given in Table 10 while Fig. 2 shows the evolution
of ethyl acetate conversion as a function of time for different M/E
ratios at the same reaction conditions. The minimum required
batch time needed to reach 95% of the equilibrium conversion is
reduced when M/E is increased. The yearly production is also
increased. However, the absolute amount of product obtained per
batch is also reduced since the initial amount of ethyl acetate is
lower at higher M/E molar ratios. This results in an increase of the
number of batches and thus an increase of the time needed for
emptying and reloading the reactor, thus resulting in an overall
productivity loss. A more in-depth productivity analysis is
therefore needed to determine the optimal M/E ratio to maximize
the profit.
The effect of using concentrations instead of activities in the
rate expressions, as discussed in Section 2.1.1, was found to be
small around the optimal process conditions, i.e. 293 K and
M/E = 10, as shown in Table 10 (compare the first entry with
the second entry). At M/E > 1, the activity coefficients of
methanol and ethanol are close to 1.0 and only a small influence
can thus be expected.
A comparison has been made with homogeneously
catalyzed transesterification. Schmidt et al. have proposed a

Table 8
Influence of the temperature on the annual production (simulation performed using Eqs. (3) and (5) at M/E = 10, W = 2 kg, VL = 10 m3, XEtOAc = 90%)
Temperature (K)

Minimum required batch time (h)

MeOAc production (tonnes year1)

batches year1

293
303
313
323

51.1
39.6
31.8
25.6

233.5
301.3
375.1
465.3

171.3
221
275.2
341.3

T.F. Dossin et al. / Applied Catalysis B: Environmental 67 (2006) 136148

145

conditions as the heterogeneous catalyzed industrial transesterification, i.e. 293 K, M/E = 10, VL = 10 m3 integrating
Eq. (5) using Eq. (25) as rate equation. In order to achieve
the same yearly production of 500 tonnes year1, 0.150 mol/m3
catalyst concentration is needed, i.e. 0.170 kg of potassium
butanolate catalyst for a comparable batch time and a 10 m3
reactor volume. The required mass of homogeneous catalyst is
thus almost 30 times smaller than the mass of solid magnesium
oxide catalyst to achieve the same production. However,
the lower cost of the solid magnesium oxide catalyst, the
elimination of the reaction mixture neutralization step and
the easy catalyst separation from the reaction mixture make
the heterogeneous process a promising alternative to the
homogeneous process.
Fig. 1. Conversion of ethyl acetate vs. batch time at different temperatures, at
M/E = 10, W = 2 kg, VL = 10 m3: () 293 K; (  ) 303 K; ( ) 313 K;
( - -) 323 K. The simulated values were obtained from Eqs. (3) and (5)
using the parameter values of Table 3.

kinetic model for the transesterification of butyl acetate with


ethanol catalyzed by potassium butanolate to form ethyl acetate
and butanol [22]. The activation energy and the pre-exponential
factor of the rate coefficient were derived from experimental
data, based on a second-order rate law (Eq. (25)) and
accounting for the polarity of the reaction mixture:


CBuOH CEtOAc
rBuOK kBuOK CBuOAc CEtOH 
Keq

(25)

This model has been applied to the homogeneous catalyzed


transesterification of ethyl acetate with methanol using the
kinetic parameters reported by Schmidt et al.: activation energy
and pre-exponential factor amount to 51 kJ/mol and 24.4 m6/
mol2 s, respectively [22]. Note that this activation energy is
twice that for the heterogeneously catalyzed reaction used in
this paper. A simulation has been performed at the same process

3.2. Continuous slurry reactor for biodiesel production


Simulations were performed of the industrial scale biodiesel
production in a continuously perfectly mixed slurry reactor by
transesterification of triolein with methanol using an heterogeneous MgO catalyst. The influence of the temperature, the
space time and the MeOH/triolein feed molar ratio (M/T ratio)
on the production has been investigated to achieve about
100,000 tonnes of methyl oleate per year. As discussed in
Section 2.2.3, the results are to be interpreted in a qualitative way
only, since the equilibrium conversion to the desired product
methyl oleate is much lower than the ones found experimentally.
In order to obtain a high productivity, the amount of catalyst
was chosen quite high: 5700 kg. In a total liquid volume of
25 m3, the volume fraction of magnesium oxide catalyst
amounts to about 0.091. According to Eq. (4), with this catalyst
fraction, the impeller revolution speed should be at least
0.76 s1 at the standard conditions: 323 K, a M/T molar ratio of
10, a space time of 1000 kg-cat s/mol-triolein, and a catalyst
particle size of 25 mm. Therefore, an impellor speed of 1.0 s1
seems sufficient. However, impeller revolutions of 5 and 10 s1

Table 9
Influence of the catalyst mass on the annual production (simulation performed using Eqs. (3) and (5) at T = 293 K, M/E = 10, VL = 10 m3, XEtOAc = 90%)
W (kg)

Minimum required batch time (h)

MeOAc production (tonnes year1)

Batches year1

2
4.5
5
10
15

51.1
22.7
20.4
10.2
6.8

233.5
525.5
583.9
1167.8
1751.6

171.3
385.4
428.2
856.5
1284.7

Table 10
Influence of the M/E molar ratio on the annual production (simulation performed using Eqs. (3) and (5) at T = 293 K, W = 5 kg, VL = 10 m3, XEtOAc = 95% of Xeq)
M/E

Minimum required batch time (h)

MeOAc production (tonnes year1)

Batches year1

10
10 a
15
20
30

20.4
19.9a
9.9
5.7
2.8

583.9
601.4a
873.8
1163.3
1637.9

428.2
441.1a
883.3
1545.8
3122.1

Simulations using rate Eq. (3a) with concentrations instead of activities.

146

T.F. Dossin et al. / Applied Catalysis B: Environmental 67 (2006) 136148

Fig. 2. Conversion of ethyl acetate vs. batch time at different M/E ratios, at
T = 293 K, W = 5 kg, VL = 10 m3: () M/E = 10; (  ) 15; ( ) 20; ( - -)
30. The simulated values have been obtained using Eqs. (3) and (5). The
simulated values were obtained from Eqs. (3) and (5) using the parameter values
of Table 3.

Fig. 4. Yields of diolein ( ), monoolein ( - -), glycerine (  ), and


methyl oleate () as a function of space time (kgcat s/mol-triolein) at
323 K, M/T = 10, W = 5700 kg, VL = 25 m3, NI = 5 s1, dp = 25 mm. The simulated values were obtained using Eqs. (11a)(11c), (13), (15), and (20), and
using the kinetic parameter values from Table 3.

were chosen in order to achieve a homogeneous mixture (or a


micro-emulsion) of the methanol and glycerides.
The reaction rates were calculated using Eqs. (12a)(12c).
The conversion of the triolein, together with the yearly
production of methyl oleate, is plotted in Fig. 3 as a function of
space time at the standard set of reactor conditions: M/T feed
ratio = 10, T = 323 K, dp = 25 mm, VL = 25 m3, triolein feed
molar flow = 5.0 mol s1, Wcat = 5700 kg, NI = 5 s1. As
expected, the conversion increases with space time until
equilibrium conversion is reached. The yearly production of
methyl oleate decreases strongly with increasing space time
due to the decrease of the reaction rate upon approaching the
equilibrium composition. Working at short space times will
thus result in a high production; however, the quality of the
biodiesel will be worse due to the larger amounts of unreacted
triolein and diolein. This can be seen in Fig. 4 in which the
yields of the products are plotted as a function of space time.
For the explorative simulations discussed in this paper, it was
assumed that a production of 100,000 tonnes methyl oleate per

reactor per year is typical. Additionally, it was assumed that the


minimum required triolein conversion and methyl oleate yields
are close to the values shown in Table 11. Of course, a thorough
economic evaluation is needed to determine the real optimum
operation conditions, which includes the maximum allowed
contents of remaining glycerides in the biodiesel according to
the ASTM or DIN standards, and also the value of the
byproduct glycerine. Currently glycerine is still a valuable
product that can be used in several industrial applications;
however, its value will decrease once the biodiesel production
takes off on a large scale.
The influence of temperature, M/T ratio, particle diameter,
and impeller revolution speed on conversions and yields were
investigated and some of these results are presented in Table 11.
According to the simulations, a production of 100,000 tonnes
methyl oleate is achieved at a space time of 871 kg s/moltriolein at 323 K, M/T feed ratio = 10, dp = 25 mm and
NI = 5 s1. A triolein conversion of 93% is achieved while
methyl oleate and glycerine yields amount to 0.54 and 0.05,
respectively. At the same conditions, according to Eq. (24),
944 tonnes of glycerine per year are produced. Separation from
methyl oleate is easy since glycerine is practically immiscible
with methyl oleate and therefore can be recovered by settling
(decantation) [7] or centrifugation [46].
Increasing the M/T ratio from 10 to 20 results in a significantly
higher triolein conversion and also higher yields of methyl oleate
and glycerine (compare first entry with the second entry in
Table 11) as was also pointed out by Boocock [46]. These yields
rise from 0.54 and 0.05 to 0.62 and 0.10, respectively. However,
this increase of M/T ratio also increases methanol flow, leading to
a larger effort to recycle the methanol, as pointed out by
Connemann et al. [7] and Zhang et al. [8]. The extra cost for
separating the methanol from the reaction mixture by means of
distillation will thus compensate the advantage of the higher
yields of methyl oleate and glycerine. An economic evaluation
will thus be needed to estimate the optimal M/T ratio.
It appears that the influence of the temperature, as well as
that of the catalyst particle size on the reactor performance is

Fig. 3. Conversion of triolein (; left axis) and the yearly production (kilotonnes year1) of methyl oleate ( ; right axis) and glycerine (  ; right axis)
as a function of space time (kgcat s/mol-triolein) at 323 K, M/T = 10,
W = 5700 kg, VL = 25 m3, NI = 5 s1, dp = 25 mm. The simulated values were
obtained using Eqs. (11a)(11c), (13), (15), and (20), and using the kinetic
parameter values from Table 3.

T.F. Dossin et al. / Applied Catalysis B: Environmental 67 (2006) 136148

147

Table 11
Space time, yields, and conversions obtained at a production rate of 100,000 tonnes methyl oleate per year as a function of the methanol/triolein feed ratio and the use
of concentrations instead of activities in the rate equations (Eqs. (12a)(12c)) instead of Eqs. (11a)(11c))
T (K)

M/T

dp (mm)

NI (s1)

W/Ftot (kg s/mol)

YMeOl

YG

YD

YM

XT

XMeOH

323
323
323

10
20
10

25
25
25

5
5
5

870.6
1013.5
851.6

0.54270
0.62279
0.52718a

0.05002
0.10319
0.04366a

0.28953
0.18789
0.30708a

0.59090
0.68123
0.56878a

0.92938
0.97137
0.91843a

0.16215
0.09300
0.15757a

Simulations using Eqs. (12a)(12c) with concentrations instead of activities.

very small. This is due to the rather close approach to


equilibrium, which is neither influenced by the temperature,
nor by the particle size. However, since a rather strong diffusion
limitation of triolein occurs at low conversion, the use of smaller
catalyst particles will significantly increase the initial conversion
rate whereas the temperature has only a minor effect. The strong
diffusion limitation of triolein at low conversion is due to the
combination of the relatively fast reaction rate and the fact that
the reaction rate is zeroth order with respect to triolein. This
zeroth order causes the rate to remain high if the triolein
concentration already decreases due to the diffusion limitation.
Obviously, both external and internal transport limitations
disappear when equilibrium is approached.
Also the impeller revolution speed has very little influence
on the reactor performance since the conversion rate is not
significantly influenced by external mass transfer limitation
(i.e. transfer from the liquid bulk towards the catalyst particles).
Nevertheless, it remains crucial to check whether the stirring
rate is really sufficient to achieve a single-phase system or
micro-emulsion system and a good suspension of the catalyst
particles in the reactor.
The effect of using concentrations instead of activities in the
rate expressions, as discussed in Section 2.2.2 was found to be
small at all conditions considered. At the standard conditions the
calculated effect is shown in Table 11 (compare the first entry
with the third entry). Apparently, the net effect of the activity
coefficients on the value of the denominators of the rate equations
is small. The numerator of the rate equations hardly changes
since the activity coefficient of methanol is always close to 1.0.
As for the EtOAc/MeOH-case, the biodiesel production
catalyzed by homogeneous catalysts, based on data from Zhang
et al. [8] and Connemann et al. [45], has been compared to the
heterogeneously catalyzed transesterification. According to
Zhang et al., a stream of 10 kg/h of sodium hydroxide is needed
to reach a production of 8000 tonnes biodiesel year1 at 333 K
with M/E = 6. This leads to a yearly catalyst consumption of 88
tonnes. Based on Connemanns data on existing biodiesel
plants, which typically operate at 333348 K, 34 kg catalyst
per tonne of biodiesel is needed to reach a yearly production of
80,000125,000 tonnes of biodiesel. This thus requires more
than 300 tonnes of homogeneous base catalyst. These amounts
are much larger than the estimated amount of 5.7 tonnes of
heterogeneous MgO catalyst that follows from the simulations
in this paper. In contrast to the batch reactor simulation for the
EtOAc/MeOH-case, the heterogeneous catalyst thus seems to
be much more efficient for biodiesel production than the
homogeneous catalysts. Although this is an encouragement to
intensify the research on the heterogeneously catalyzed system,

it must be interpreted with care in view of the assumptions and


simplifications made in the explorative simulations presented
here. Particularly the assumption of the reaction kinetics to be
exactly the same as for the transesterification of ethyl acetate
may have a major impact. On the other hand it might be that the
catalytic activity of the KOH catalyst in the homogeneous
process is rather low. Despite the uncertainties in the
explorative simulation results of the biodiesel production
process, they offer a useful starting point towards further
investigation of the industrial biodiesel production using MgO
catalysts. Moreover, these simulation results indicate that
heterogeneous catalysts may allow good production levels and
present an interesting alternative to homogeneous catalysts
towards a more environment-friendly process.
4. Conclusions
Magnesium oxide offers a viable heterogeneous solid base
catalyst for the standard transesterification of ethyl acetate or
triolein with methanol at industrial conditions for the industrial
production of fine-chemicals or biodiesel. Simulation of batch
and continuous slurry reactors, based on an EleyRideal kinetic
model, show that industrial production of fine-chemicals can be
reached at ambient temperature and atmospheric pressure using
finely dispersed MgO catalyst. This allows the development of
a more environmentally friendly chemical process for the
transesterification reaction of low and high methyl acetate
production volumes since reaction neutralization is avoided and
product separation is made much easier. Also, the use of MgO
makes the batch process cost-effective since reactor heating is
not required.
Explorative process simulations for the continuous production of biodiesel indicate that commercially attractive production rates can be achieved. However, further experimental
investigation is needed to quantitatively determine the potential
of MgO as an alternative towards a more sustainable
development of the industrial biodiesel production process.
Acknowledgement
This research has been funded in the frame of the STWW
project HYPCAT (STWW 141) of the IWT of the Flemish
Government.
References
[1] F. Ma, M.A. Hanna, Biodiesel, Bioresour. Technol. 70 (1999) 1.
[2] J. Otera, Chem. Rev. 93 (1993) 1449.

148

T.F. Dossin et al. / Applied Catalysis B: Environmental 67 (2006) 136148

[3] U. Meyer, W.F. Hoelderich, Appl. Catal. A 178 (1999) 159.


[4] A. Corma, S. Iborra, S. Miquel, J. Primo, J. Catal. 173 (1998) 315.
[5] J. Barrault, Y. Pouilloux, J.M. Clacens, C. Vanhove, S. Bancquart, Catal.
Today 75 (2002) 177.
[6] D. Darnoko, M. Cheryan, J. Am. Oil Chem. Soc. 77 (12) (2000) 1269.
[7] J. Connemann, A. Krallmann, E. Fischer, Process for the continuous
production of lower alkyl esters or higher fatty acids, US Patent 5354878,
1993.
[8] Y. Zhang, M.A. Dube, D.D. McLean, M. Kates, Bioresour. Technol. 89
(2003) 1.
[9] N. Ergun, P. Panning, Method for producing fatty acid methyl esters and
equipment for realizing the same, US Patent 6440057, 2002.
[10] M. Canonge, J.-C. Joly, Process for the continuous preparation of acetates,
Patent EP0280471A1, 1988.
[11] A.F. Eastgate Sims, J. Southwood, Production of acrylic and methacrylic
esters, Patent GB960005, 1961.
[12] Y.-J. Fang, W.-D. Xiao, Sep. Purif. Technol. 34 (2004) 255.
[13] L. Jimenez, A. Garvin, J. Costa-Lopez, Ind. Eng. Chem. Res. 41 (2002)
6663.
[14] B. Davies, G.V. Jeffreys, Trans. Inst. Chem. Eng. 51 (1973) 271.
[15] B. Davies, G.V. Jeffreys, Trans. Inst. Chem. Eng. 51 (1973) 275.
[16] B. Freedman, R.O. Butterfield, E.H. Pryde, J. Am. Oil Chem. 63 (10)
(1986) 1375.
[17] S. Gryglewicz, Appl. Catal. A 192 (2000) 2328.
[18] H. Hattori, M. Shima, H. Kabashima, Stud. Surf. Sci. Catal. 130 (2000)
3507.
[19] M. Di Serio, R. Tesser, A. Ferrara, E. Santacesaria, J. Mol. Catal. A (2004)
251.
[20] U. Schuchardt, R. Sercheli, R.M. Vargas, J. Braz. Chem. Soc. 9 (1) (1998)
199.
[21] L. Bournay, D. Casanave, B. Delfort, G. Hillion, J.A. Chodorge, Catal.
Today 106 (2005) 190.
[22] J. Schmidt, D. Reusch, K. Elgeti, R. Schomacker, Chem. Ing. Tech. 71
(1999) 704.
[23] H. Noureddini, D. Zhu, J. Am. Oil Chem. Soc. 74 (11) (1997) 1457.
[24] D. Darnoko, M. Cheryan, J. Am. Oil Chem. Soc. 77 (12) (2000) 1263.
[25] T.F. Dossin, M.-F. Reyniers, G.B.M. Marin, Appl. Catal. B Environ. 61
(2006) 35.
[26] A. Demirbas, Energy Conversion Manage. 44 (2003) 2093.
[27] ASPEN Engineering SuiteTM 11.1 version, Aspen Technology Inc.

[28] C. Reid, J.M. Prausnitz, D.E. Poling, The Properties of Gases and Liquids,
third and fourth ed., McGraw-Hill, New York, 1987.
[29] R.J. Madon, I. Iglesia, J. Mol. Catal. A Chem. 163 (2000) 189.
[30] G.B. Tatterson, Fluid Mixing and Gas Dispersions in Agitated Tanks,
McGraw-Hill, New York, 1991.
[31] D. Thoenes, Chemical Reactor Development, Kluwer Academic Publishers, Dordrecht, The Netherlands, 1994, p. 91.
[32] F. Kapteijn, G.B. Marin, J.A. Moulijn, Catalytic Reaction Engineering: An
Integrated Approach of Homogeneous, Heterogeneous and Industrial
Catalysis, Elsevier, 1993 (Chapter 7 in Catalysis).
[33] G.F. Froment, K. Bischoff, Chemical Reactor Analysis and Design, J.
Wiley & sons, 1990.
[34] D.E. Mears, J. Catal. 20 (1971) 127.
[35] H. Dunski, W.K. Jozwiak, H. Sugier, J. Catal. 146 (1994) 166.1.
[36] B.C. Gates, Catalytic Chemistry, Wiley, New York, 1992.
[37] L.R. Petzold, SIAM J. Stat. Comput. 4 (1983) 136.
[38] Netlib, http://www.netlib.org.
[39] J.M. Encinar, J.F. Gonzalez, A. Rodriguez-Reinares, Ind. Eng. Chem. Res.
44 (2005) 5491.
[40] Y. Sano, N. Yamaguchi, T. Adachi, J. Chem. Eng. Jpn. 7 (1974) 255.
[41] F. Kapteijn, J.A. Moulijn, in: G. Ertl, H. Knozinger, J. Weitkamp (Eds.),
Laboratory Reactors: Handbook of Heterogeneous Catalysis, vol. 3, VCH
Verlagsgesellschaft mbH, Weinheim, Germany, 1997, p. 1359 (Chapter 9).
[42] C.S. Caldwell, A.L. Babb, J. Phys. Chem. 60 (1956) 51.
[43] R.H. Perry, D.W. Green, J.O. Maloney, Perrys Chemical Engineers
Handbook, seventh ed., McGraw-Hill, New York, 1997(Chapter 5, pp. 552).
[44] D. Bockey, Situation and Development Potential for the Production of
BiodieselAn International Study, Union zur Forderung von Oel und
Proteinpflanzen e.V., 2002.
[45] J. Conneman, J. Fischer, International Liquid Biofuels Congress, Curitiba,
Brazil, July 1922, 1998.
[46] D.G.B. Boocock, Process for productivity of fatty acid methyl esters from
fatty acid triglycerides, US Patent No. 6,712,867 B1, March 30, 2004.
[47] X. Lang, A.K. Dalai, N.N. Bakhshi, M.J. Reaney, P.B. Hertz, Bioresour.
Technol. 80 (2001) 53.
[48] G. Knothe, K.R. Steidley, Fuel 84 (2005) 1059.
[49] http://www.patechfc.com.tw/p2e.htm, November 2005.
[50] G. Knothe, Fuel Proc. Technol. 86 (2005) 1059.
[51] C.W. Weast, CRC Handbook of Chemistry and Physics, 63rd ed., CRC
Press, Boca Raton, USA, 19821983.

You might also like