Vous êtes sur la page 1sur 14

Chemical Engineering Journal xxx (2012) xxx–xxx

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Journal


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

Simulation and kinetic study of transesterification of triolein to biodiesel


using modular reactors
J.F. Portha a,⇑, F. Allain a, V. Coupard b, A. Dandeu b, E. Girot a, E. Schaer a, L. Falk a
a
Université de Lorraine, Laboratoire Réactions et Génie des Procédés, UPR CNRS 3349, Nancy F 54001, France
b
IFP Energies Nouvelles, Solaize F 69360, France

h i g h l i g h t s

" Transesterification of triolein to produced biodiesel is considered.


" A two stages continuous heterogeneous catalytic reactor is used.
" A 2D model has been developed for the understanding of coupled phenomena.
" Kinetic parameters have been determined.
" Staging is interesting when strong mass transfer limitations occur.

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

Article history: A laboratory pilot unit has been developed to solve the issues of thermodynamic equilibrium shift for the
Available online xxxx transesterification of vegetable oil into biodiesel. It consists in a two stages continuous heterogeneous
catalytic reactor operating at 50 bar and 175 °C. Under the working conditions, only one homogeneous
Keywords: phase can be assumed. The reactions of transesterification need a large excess of methanol implying
Heterogeneous catalytic reactors an additional energy consumption and higher capital expenditure because of larger size of equipments,
Intensification and operating pressure.
Series and parallel reversible reactions
In this work, a numerical model has been developed in order to better understand the different coupled
Transesterification of biodiesel
Finite volume method
phenomena (thermodynamic equilibrium, kinetics, mass transfer limitations, etc.) governing biodiesel
production. The model is based on a heterogeneous plug flow reactor with axial dispersion taking into
account mass transfer in the catalyst and dynamic aspects. The reactor is supposed isothermal since
transesterification reactions are almost athermic.
Kinetic parameters have been optimized by comparison with experimental results previously obtained
by IFPEn. Simulation results are used to establish the future experimental conditions that will be used on
the pilot unit. Mass transfer limitations are characterized by changing the diameter of catalyst particle.
Concentration and reaction rate profiles are depicted in the fluid phase and in the catalyst. The simula-
tions indicate that a conversion of triolein of 87% can be reached with a methanol to triolein molar ratio of
36 at reactor inlet and with a mean residence time of 1 h.
Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction conventional diesel fuel. Biodiesel is produced from vegetable oils


(mainly rapeseed, soybean or sunflower oil) by transesterification
The production of biodiesel, a first generation biofuel, has been to reduce viscosity of vegetable oil. This fuel can be used directly
strongly developed in some countries as a substitute for in diesel engines but is commonly mixed with petroleum-based
diesel.
Abbreviations: IFPEn, IFP Energies Nouvelles; T, triglyceride or triolein; D,
The main advantages of biodiesel are its ability to reduce the
diglyceride; M, monoglyceride; E, ester or methyl oleate; MeOH, methanol; W, climate change impact (limitation of greenhouse gas emissions
water; G, glycerol; MeG, methoxyglycerol; DGE, diglycerolether. due to carbon dioxide consumption by the plants during their
⇑ Corresponding author. Address: LRGP, 1 rue Grandville, 54001 Nancy, France. growing), its renewability, non-toxicity, and the reduction of car-
Tel.: +33 383 17 53 70.
bon monoxide, sulfur compound and particulate matter emission
E-mail addresses: jean-francois.portha@ensic.inpl-nancy.fr, jean-francois.
portha@univ-lorraine.fr (J.F. Portha).
when compared to conventional diesel fuel. Moreover, biodiesel

1385-8947/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cej.2012.06.106

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
2 J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx

Nomenclature

Latin symbols el bed (external) porosity (–)


1 1
Aj pre-exponential factor of reaction j ðm6 mol s1 kgcata Þ ep catalyst (internal) porosity (–)
aLS liquid/solid surface to volume ratio (m2 m3) ui thiele modulus related to component i
Ci concentration of component i (mol m3) m = u/e1 liquid interstitial velocity (m.s1)
C li concentration of component i in the liquid (mol m3) mij stoichiometric coefficient of component i in
C si concentration of component i in the solid (mol m3) reaction j (–)
DA axial dispersion (m2 s1) Ui association factor of component i (–)
Dm diffusivity (m2 s1) gi catalyst efficiency related to component i (–)
Di,j diffusivity of component i in component j (m2 s1) gi dynamic viscosity of component i (cP)
Deff,i,j effective diffusivity of component i in component j qp catalyst density (kg.m3)
(m2 s1) s residence time (s)
dp catalyst particle diameter (m) sg global residence time (s)
dt reactor diameter (m) sp catalyst tortuosity (–)
EA activation energy of reaction j (J mol1) XR reactor section (m2)
Fi molar flow rate of component i (mol s1) XL reactor section occupied by the liquid (m2)
f shape factor (-)
Kj thermodynamic constant of reaction j (-) Dimensionless numbers
kLS liquid/solid mass transfer coefficient (m s1) PeA ¼
mdp
axial Peclet number
DA
L characteristic length (m)
q ud
Mi molar mass of component i (g mol1) Re ¼ ll p particle Reynolds number
l
P pressure (Pa) Sc ¼ Dm schmidt number
flowing volume (m3 s1)
m
Q
1 Sh ¼ kDLSmL sherwood number
rj rate of reaction j ðmol s1 kgcata Þ
P 1
ri ¼ j mij r j formation rate of component i ðmol s1 kgcata Þ
Subscripts
r radial position in catalyst particle (m)
T triolein or triglyceride
Rg ideal gas constant (J mol1 K1)
D diglyceride
Rp catalyst particle radius (m)
M monoglyceride
R inlet methanol/triolein molar ratio
MeOH methanol
S = 4pr2 catalyst surface (m2)
E ester (methyl oleate)
T temperature (K or °C)
G glycerol
t time (s)
n, N North
u liquid superficial velocity (m s1)
s, S South
VL liquid volume (m3)
E East
VR reactor volume (m3)
W West
vi molar volume of component i (cm3 mol1)
P central point
X conversion (-)
xi molar composition of component i (-)
Superscripts
Y yield (-)
e inlet reactor
z axial position (m)
s related to the solid phase
zf reactor length (m)
l related to the liquid phase
Greek symbols
a feed ratio (–)

addition in conventional diesel fuel, at levels of 1 – 2% has the conditions has been studied by some authors [10] to solve purifica-
beneficial effect of restoring lubricity lost due to the restricted sul- tion and separation difficulties. Optimal conditions were shown to
fur content. The drawbacks are that biodiesel emits more NOX than be 350 °C, 190 bar and a molar methanol to oil ratio of 42. The
conventional fuel and is competing with food production for land supercritical state of methanol is interesting because only one sin-
use. This last point supports the recent development of second gle phase is formed.
and third generation biofuels. Otherwise, the use of heterogeneous catalyst may be a solu-
Biodiesel consists of alkyl esters of fatty acids produced by tion because it can be easily separated from the liquid mixture.
transesterification of vegetable oil by methanol or ethanol. Vegeta- Kim et al. [9] have shown that the use of a heterogeneous base
ble oils are triglycerides. In this paper, the chosen model reaction is catalyst (Na/NaOH/c-Al2O3) in a stirred batch reactor presents
the transesterification of trioleine (C57H104O6) with methanol to almost the same activity under optimal conditions (conversion
produce methyl oleate (CH3(CH2)7CH@CH(CH2)7COOCH3) and glyc- of 95%, methanol/oil ratio = 9, n-hexane as co-solvent, atmo-
erin over a solid alkaline catalyst. The density of methyl oleate is spheric pressure) compared to the conventional homogeneous
870 kg m3 at 298 K. catalyst (NaOH). The performance of alkaline catalysts is very
Transesterification reactions are usually performed with dependent on the methanol/oil molar ratio because an increase
homogenous base or acid catalysts which are soluble in methanol. of this ratio implies an increased driving force for methanol
This method presents however some drawbacks because neutral- adsorption [1].
ization and separation of the catalyst is required. Other methods Some studies focus on thermal non-catalytic transesterification
are described in the literature: The use of supercritical methanol of oil in batch reactor [5] or on the opportunity to convert triglyc-
for transesterification without catalyst under different reaction erides in a wide range of raw materials [2].

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx 3

The transesterification reaction is promoted by high tempera- Moreover, two non-balanced side reactions take place (second-
tures (to enhance kinetics) and pressures that increase the solubil- ary mechanism):
ity of methanol in oil. Moreover, catalysts operating at high
temperature and pressure are very stable and unlike many other G þ MeOH ! MeG þ W ðR4Þ
heterogeneous catalysts, neither deactivation, nor leaching of ac-
tive phase in the glycerine or ester has been observed [12]. 2G ! DGE þ W ðR5Þ
The sizes of tested catalyst particle are 2.5 mm and 5 mm. This
quite large particle size is interesting from two points of view.
In the first reaction, triglyceride (or triolein) (T) reacts with
Firstly, pressure drops in the plug flow reactor are minimized. A
methanol (MeOH) to produce diglyceride (D) and methyl oleate
large decrease of pressure, promoted by small particle size, may
(E), a fatty acid ester. Then, in the second reaction, diglyceride re-
imply local vaporization of methanol (the vapor pressure of meth-
acts with methanol to form monoglyceride (M) and another mole-
anol at 175 °C is 25 bars); consequently, the viscosity of the mix
cule of methyl oleate. In the third reaction, monoglyceride reacts
may be increased leading to reactor fill-in.
with methanol to produce glycerol (G) and methyl oleate. Side
Secondly, the use of a fine powder leads to leaching of the cat-
reactions imply two reactions with glycerol producing respectively
alyst. This phenomenon was not observed by using extrudates
methoxyglycerol (MeG) and water (W), and diglycerolether (DGE)
(2 mm of diameter and 3–5 mm of length). Standards are very
and water.
strict on fuels quality and especially the amount of metal impuri-
ties has to be very low. The purity of glycerine is also an important
economical parameter because pharmaceutical applications are 2.2. Kinetic model
also possible. The opportunity to use separation processes (centri-
fugation) is very expensive. Experimental data have been previously used by IFPEn to estab-
The molar ratio of methanol to oil is one of the most important lish a kinetic model based on a second-order rate law. The experi-
variables affecting the conversion. The stoichiometry of the transe- ments have been realized on an existing pilot unit. Expressions of
sterification process requires three molecules of methanol to react reaction rates are given by the following equations:
with one molecule of triglyceride but studies have shown that a  
EA
large excess of methanol is required to shift the thermodynamic R 1 1
r 1 ¼ A1  e g T  C T  C MeOH   CD  CE ð1Þ
equilibrium. This technique may involve large additional energy K1
consumption (downstream work-up and recycling) and higher
capital expenditure because of larger size of equipment. An inter- EA  
R 2 1
esting alternative consists in the development of multi-stage reac- r 2 ¼ A2  e g T  C D  C MeOH   CM  CE ð2Þ
K2
tors with inter-stage feeds that enables both the shift of
thermodynamic equilibriums and lower reactants excess. EA  
For multiphase Liquid/Liquid reactions, additional effects also R 3 1
r 3 ¼ A3  e g T  C M  C MeOH   CG  CE ð3Þ
have to be considered, such as dynamic mixing between phases, K3
mass transfer and reaction kinetics.
An industrial application has been developed by IFPEn, using R
EA
4

heterogeneous base catalysts (mixed oxide of zinc and aluminum) r 4 ¼ A4  e g T  CG ð4Þ


for the transesterification of vegetable oils to produce Biodiesel.
The process is operated in continuous mode and includes two suc- R
EA
5

cessive fixed bed reactors [3]. Authors have shown that the yield of r 5 ¼ A5  e g T  C 2G ð5Þ
methyl ester is very high and that the glycerol content of the glyc-
with Ai, the pre-exponential factor, EAi the activation energy and Ki
erin by-product is also very high (98%).
the thermodynamic equilibrium constant of reaction i. Reaction rate
The understanding of the coupled phenomena in the catalytic
expressions taking adsorption phenomena into account can be
fixed bed (mass transfer, kinetics and thermodynamic equilibrium)
found in the literature. Dossin et al. developed for example a reac-
will be improved by the development of a model and with corre-
tion mechanism of transesterification over MgO catalyst with an
sponding simulations.
Eley–Rideal type methanol adsorption as the rate determining step
In this paper, the catalytic transesterification of triolein into
[6]. This mechanism implies the reaction between an adsorbed spe-
Biodiesel is considered. Simulation result will enable to find the
cies A and a molecule B non-chemically adsorbed. In the present
optimal process configuration when a two-stage reactor is con-
work, the species adsorption is not taken into account in the kinetic
sidered. This means for example how to inject optimally the
expressions.
reactants (here methanol) at reactor inlet and how to control
temperature.
2.3. Thermodynamical assumptions
2. Presentation of the model
The reaction of transesterification of ethyl acetate with metha-
2.1. Chemical reactions nol is slightly exothermic. In the present study, the transesterifica-
tion of triglyceride is assumed to be athermic.
The transesterification of vegetable oil consists of the following At ambient conditions, the mixture consists in two different
three combined series and parallel reversible reactions (primary phases: one aqueous and one organic phase. At working conditions
mechanism): (temperature of 175 °C and pressure of 50 bar), only one phase has
been observed and this assumption will be taken into account in
T þ MeOH $ D þ E ðR1Þ the model. The mixture is non ideal. Despite this fact, concentra-
tions have been considered instead of activities in the kinetic
D þ MeOH $ M þ E ðR2Þ
expressions. However, previous studies have shown that, for high
values of methanol to ester molar ratio, only a small difference is
M þ MeOH $ G þ E ðR3Þ
observed between the use of a non ideal and an ideal model [7].

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
4 J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx

2.4. Preliminary estimation of the limiting phenomena 2.4.2. External mass transfer between the fluid and the catalyst
particle
The performances of the transesterification reactor is a result of The external mass transfer characteristic time is defined by the
the complex coupling of competitive phenomena of fluid flow, expression:
external and internal mass transfer in the catalyst, and rate of
chemical reactions controlled by the thermodynamic equilibrium. L2 L dp
tmat ¼ ¼ ¼ ð10Þ
In order to evaluate the coupling, the characteristic times of the ShDm kLS 6ð1  el ÞkLS
different phenomena have been estimated.
with aLS the specific area between the flowing fluid and the external
A non-limiting step phenomenon presents a characteristic time
surface of the particle and kLS the mass transfer coefficient. Sher-
much lower than the other ones. A slow phenomenon exhibits a
wood number is related to the particle Reynolds number and
relative larger characteristic time, implying a decrease of process
Schmidt number, with the Ranz and Levenspiel correlation [13]:
productivity. The use of characteristic times analysis may also give
information to identify which step should be accelerated in order Sh ¼ 2 þ 1:8Re0:5 Sc0:33 ð11Þ
to maximize reactor productivity or selectivity [4,8].
Since the process is assumed to be isothermal, no characteristic
time in relation with heat transfer was calculated. 2.4.3. Rates of the reactions
The characteristic time of the first reaction, calculated at reactor
inlet, is given by the expression:
2.4.1. Internal mass transfer in the catalyst particle
The diffusion characteristic time is defined by the following C eT
treac ¼ ð12Þ
equation: re
with C eT the concentration of triolein at the reactor inlet and re, the
L2
t diff ¼ f ð6Þ rate of reaction calculated with the inlet concentrations. The other
Deff reactions are not considered here, because initial concentration of
diglyceride and monoglyceride are zero.
L is the characteristic length related to the catalyst particle dimen- These characteristic times, each of them describing a different
sion (with L = dp/6 for spherical particles), Deff is the effective diffu- physical phenomenon (internal and external mass transfer, chem-
sivity. The calculation of this parameter is explained later. The ical reaction) have to be compared to check if the steady state re-
expression of the shape factor f is given by: gime is reached and if internal or external mass transfer limitations
exist.
pþ1
f ¼ ð7Þ Table 2 gives the different typical numerical values of the pilot
pþ3 plant parameters: temperature and pressure, reactor size.
LHSVOil is the liquid hourly space velocity defined according to
with p = 1 for a plate, p = 2 for a cylinder and p = 3 for a sphere. oil volume flow rate at 15 °C and is given by the following
Diffusion coefficients were evaluated for each compound with equation:
the use of the Wilke and Chang correlation. As methanol is used
in large excess, the diffusivity of species i in methanol has been Q v ;Oil
LHSV Oil ¼ ð13Þ
considered and calculated by the following expression: ð1  el ÞV R
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi The values of the Reynolds number of particle (between 0.09
UMeOH MMeOH T
Di;MeOH ¼ 7:4  108 ð8Þ and 0.3) show that the hydrodynamic regime is laminar.
gMeOH v 0:6
i

with Di,MeOH the diffusion coefficient of species i in methanol at low 2.4.4. Catalyst efficiency
concentrations (cm2 s1), UMeOH the association factor of methanol, The magnitude of the catalyst efficiency gj indicates the relative
vi the molar volume of species i at its normal boiling temperature importance of internal diffusion and reactions limitations. It can be
(cm3 mol1), MMeOH the molecular weight of methanol (g mol1), calculated, for a first order reaction in a spherical particle, by the
T the temperature (K) and gMeOH the viscosity of methanol (cP). following relationship:
We have UMeOH = 1.9, gMeOH = 0.59 cP and MMeOH = 32 g mol1. The 3ui cot hð3ui Þ  1
molar volumes vi were obtained with the additive method of Sch- gi ¼ ð14Þ
3u2i
roeder, and the temperature was fixed at 175 °C.
Effective diffusivities are then obtained with the following where ui represents the Thiele Modulus for component i. Its expres-
equation: sion is given by:
 2
Di;j ep r i qp dp
Deff ;i;j ¼ ð9Þ u2i ¼ ð15Þ
sp Deff ;i;MeOH C i 6
Tortuosity and catalyst porosity are respectively equal to The Thiele Modulus is the ratio of the intrinsic chemical reac-
sp = 2.5 and ep = 0.52. Diffusion and effective diffusion coefficients tion rate in the absence of mass transfer limitation to the rate of
values are given in Table 1. diffusion through the particle.

Table 1
Diffusion coefficient of species (m2.s1).

Species i Triolein Methanol Diglyceride Monoglyceride Glycerol Ester


Di,MeOH (m2 s1) 6.19  1010 4.29  109 7.79  1010 1.11  109 2.83  109 1.19  109
Deff,i,MeOH (m2 s1) 1.29  1010 8.92  1010 1.62  1010 2.3  1010 5.89  1010 2.48  1010

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx 5

2.5. Description of the mathematical reactor model 6


aLS ¼ ð1  el Þ ð20Þ
dp
The hydrodynamics hypotheses considered for the model is a
Axial dispersion is considered in this model. For liquid flow with
heterogeneous plug flow with axial dispersion for the continuous
a Reynolds number under 10, the axial Peclet number PeA has a va-
phase, and diffusion inside the particle catalysts for reacting spe-
lue of 0.5 [14]. As our model is assumed to be monophasic liquid,
cies. The model is unsteady and two-dimensional to take into ac-
this value was retained.
count transient regime and internal mass transfer resistance.
After some mathematical operations, the balance can be rewrit-
External mass transfer resistance is also taken into account. The
ten as:
first dimension z is the length of the reactor and the second one
r is the radius of catalyst particle. @ðuC li Þ @2 Cl 6 @C l
The reactor is assumed to be isothermal. Preliminary inquiry  DA el 2i þ kLS ð1  el ÞðC li  C si Þ þ ðel þ ð1  el Þep Þ i ¼ 0
@z @z dp @t
shows that if mixed with a proper quantity of methyl oleate, trio-
ð21Þ
lein and methanol becomes homogeneous in mixture, hereby
removing feed mixing issues.
2.5.2. Mass balance in the internal catalyst particle
2.5.1. Mass balance in the external liquid phase The balance on a component i in the catalyst particle writes as:
The balance applied to a component i in the liquid phase can be   X  
@C s @C s @C s
written as: Deff ;i S i þ mij rj dV ¼ Deff ;i S i þ i ep dV ð22Þ
! ! @r r j
@r rþdr @t
@C li @C li
mC li  DA el XR ¼ mC li  DA el XR The balance is made up of four terms: two terms of mass trans-
@z @z
z zþdz port by diffusion in the catalyst (inlet/outlet), one source-term and
  @C l one term of accumulation in the particle control volume dV. The
þ kLS aLS C li  C si dV R þ i dV L ð16Þ
@t associated boundary conditions are respectively at solid surface
and at the particle center:
The balance is made up of four terms: two terms of mass trans-
 s
port (inlet/outlet) by convection and diffusion, one term of mass @C i
transfer to the catalyst and one term of accumulation. Chemical kLS ðC li  C si Þ ¼ Deff ;i ð23Þ
@r r¼Rp
reactions take place only in the catalyst implying that no source
term is included in the balance. The associated inlet boundary con-  
@C si
dition is defined by (Dirichlet condition): Deff ;i ¼0 ð24Þ
 @r r¼0

uC li;inlet ðtÞ ¼ uC li ðtÞ ð17Þ After some mathematical operations, the balance gives:
z¼0
 s
The associated outlet boundary condition is defined by (Neu- 1 @ 2 @C i @C si X
Deff ;i r þ ep ¼ mij rj ð25Þ
mann condition): r 2 @r @r @t j
!
@C li @C li
mC li  DA el XR ¼ ðmC li Þz¼L el XR þ kLS aLS ðC li  C si ÞdV R þ dV L
@z @t 2.5.3. Mass balances discretization by the finite volume method
z
The mesh used for the resolution is represented in Fig. 1. The
ð18Þ
mass balance equations are discretized as follow: the fluid phase
The liquid volume, considering catalyst porosity ep, and bed is composed of nbz elements. Every one of these elements is con-
porosity el is calculated by: nected to a solid particle composed of nbr elements. This solid par-
ticle is assumed to be spherical and surrounded by the fluid
V L ¼ ðel þ ð1  el Þep ÞV R ð19Þ
conditions at this point and is not in contact with any other parti-
The liquid/solid surface to volume ratio is calculated as follow, cles. Therefore, to take into account that there is no mass transfer
under the assumption of spherical particles: between two different particles, no transfer is modeled between

Fluid phase

Solid phase

Fig. 1. Mesh grid for the modeling by the finite volume method.

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
6 J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx

where the source term So, an and as are defined by:


z
X Dt
SoðC si;P ðtÞÞ ¼ mij rj ðC si;P ðtÞÞ ð34Þ
N ep
W wx P ex E j
r
n Sn Dt Ss Dt
an ¼ Deff as ¼ Deff ð35Þ
P V p ep Dr V p ep Dr
s No linearization of the source-term is performed. The concen-
S trations used to calculate this term are taken at the preceding time
step t instead of t + Dt.
The associated boundaries conditions are defined by:
Fig. 2. Control volume method in the liquid and in the catalyst. !
C si;S ðt þ DtÞ  C si;P ðt þ DtÞ 
kLS ðC li;P ðtÞ  C si;P ðt þ DtÞÞ ¼ Deff ;i 
Dr 
r¼Rp
two contiguous elements in the solid phase (represented in the
horizontal direction in Fig. 1). Mass transfer by diffusion is taken ð36Þ
into account in the particle by a balance according to r (vertical 

direction in Fig. 1). The center of each solid particle is located at C si;P ðt þ DtÞ ¼ C si;S ðt þ DtÞ ð37Þ
r¼0
the bottom of the figure. As central symmetry is assumed, the vol-
ume of last element is only half the one of the standard elements. The solution of the discretized equations can be obtained by the
The mass balance in the fluid phase is integrated at step time t standard Gaussian elimination method. Because of the simple form
and each mass balance in each particle of the solid phase is then of the equations, the matrices of coefficients are tridiagonal and
solved at step time t + Dt. the TDMA algorithm can be used. The calculation are performed
Applying the finite volume method, with implicit scheme, to the with MatlabÓ software.
external mass balance equation leads to the following discretiza- The courant Friedrich Levy conditions (CFL conditions) have
tion expression [11]: (see also Fig. 2) been checked although the calculation is not required for implicit
scheme. Mass balances have been checked for each chemical
C li;W ðt þ DtÞ þ aðC li;W ðt þ DtÞ  C li;P ðt þ DtÞÞ element.

¼ C li;P ðt þ DtÞ þ aðC li;P ðt þ DtÞ  C li;E ðt þ DtÞÞ þ bðC li;P ðt þ DtÞ
3. Results for a single reactor
 C si;P ðtÞÞ þ cðC li;P ðt þ DtÞ  C li;P ðtÞÞ ð26Þ
3.1. Adjustment of the kinetics parameters
where
Determination of the kinetic parameters was conducted with
DA el
a¼ ¼ Pe1
A dp =Dz ð27Þ the use of MatlabÓ (fmincon optimization module). Thirteen
uDz
experimental points were obtained on the IFPEn pilot unit
kLS aLS V R s (Table 2), using one reactor while varying temperature, liquid
b¼ ¼ ð28Þ hourly space velocity of oil (LHSVOil) and methanol to triolein inlet
uXR t mat
molar ratio. The catalyst particle diameter is fixed to 2.5 mm. Con-
ðel þ ð1  el Þep ÞV R dition of this testing is specific to kinetic measurement only.
c¼ ð29Þ Neglecting the side reactions (reactions (R4) and (R5)), the opti-
uXR Dt
mization was carried out on the three pre-exponential constants
After some mathematical operations, the previous equation can k1, k2 and k3 and three activation energies EA1, EA2 and EA3.
be rewritten as follow: The following criterion was minimized, comparing the experi-
a   mental values of the molar compositions xexp,i with the one
2a þ b þ c þ 1
C li;P ðtÞ ¼ C li;E ðt þ DtÞ þ C li;P ðt þ DtÞ obtained with the model xcalc,i:
c c
   
a  1 b X ðxexp;i  xcalc;i Þ2
þ C li;W ðt þ DtÞ  C si;P ðtÞ ð30Þ J¼ ð38Þ
c c x2exp;i
i

The associated boundary conditions are defined by (respectively The results of this optimization are shown in Table 3. Pre-expo-
for inlet and outlet): nential factors and activation energies are given for the three main
C li;P ðtÞ ¼ C l;e ð31Þ reactions. A calculation of the kinetic constant of each reaction by
i

a   Table 2
2a þ b þ c þ 1
C li;P ðtÞ ¼ C li;E ðt þ DtÞ þ C li;P ðt þ DtÞ Range of process parameters.
c c
    Conditions
a  1 b T (°C) 150–175
 C li;W ðt þ DtÞ  C si;P ðtÞ ð32Þ P (bar) 50
c c
Methanol/oil molar ratio 1–32.75
The discretization of internal mass balance equation by the fi- LHSVOil (h1) 1–3
nite volume method, with implicit scheme, leads to the following s(s) 734–2202

expression [11]: Reactor


dt(m) 0.015
C si;P ðtÞ ¼  SoðC si;P ðtÞÞ  C si;N ðt þ DtÞ an þ C si;P ðt þ DtÞð1 þ an þ as Þ zf(m) 0.12
el 0.4
 C si;S ðt þ DtÞ as ð33Þ

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx 7

Table 3 the Arrhenius law points out that reaction (R1) is the faster one.
Optimization results on the kinetics parameters. Both reactions (R2) and (R3) are about one order of magnitude
A1 (m6 mol1 s1 kgcata-1) 9.649  104 lower than reaction (R1).
A2 (m6 mol1 s1 kgcata-1) 5.381  107 Figs. 3 and 4 show a comparison between the results of the
A3 (m6 mol1 s1 kgcata-1) 5.808  108 model and experimental points by plotting the molar composition
EA1 (J mol1) 48981
EA2 (J mol1) 28537
at the outlet of the system as a function of LHSVOil.
EA3 (J mol1) 21569 The model with the optimized parameters described perfectly
the experimental points of methanol and ester. For trioleine,

0.02

0.018

0.016

0.014
Compositions

0.012

0.01

0.008

0.006

0.004

0.002

0
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3

LHSV ( h−1)
oil

Fig. 3. Comparison between model and experiments – Molar ratios (mol/mol) —— Triolein (model), h Triolein (experimental data), – – Diglyceride (model), s Diglyceride
(experimental data), – - – - Monoglyceride (model), e Monoglyceride (experimental data) (R = 32.75, dp = 2.5 mm).

0.9

0.8

0.7

0.6
Compositions

0.5

0.4

0.3

0.2

0.1

0
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
LHSV (h−1)
oil

Fig. 4. Comparison between model and experiments – Molar ratios (mol/mol) —— Methanol (model), h Methanol (experimental data), - - - Ester (model), s Ester
(experimental data) (R = 32.75, dp = 2.5 mm).

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
8 J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx

Table 4 FT
XT ¼ 1  ð40Þ
Numerical values of characteristic times (R = 32.75). F eT
dp = 2.5 mm dp = 0.7 mm
XT 0.93 0.98 FE
YE ¼ ð41Þ
tdiff(s) 1350 100 3F eT
tmat(s) 270 30
treac(s) 40 40 FT and F eT are the molar flow rates of triolein respectively at reactor
s(s) 3600
outlet and inlet; FE is the molar flow rate of ester at reactor outlet.
Conversion and yield obtained with the model and the opti-
mized kinetics parameters are also plotted in Fig. 5 against the
diglyceride and monoglyceride, the experimental error may be rel- methanol to triolein inlet ratio and results are compared to the
atively large (10%); the accuracy of the calculated compositions is thermodynamic equilibrium.
very satisfactory and the slope of the theoretical curve is in agree- As can be seen in Fig. 5, as expected, increasing the inlet molar
ment with the slope of experimental points. As expected, an in- ratio improves the equilibrium conversion to a value close to 1 and
crease of LHSVOil, meaning a decrease of mean residence time, the equilibrium yield to a value close to 0.95. The conversion and
implies a lower triolein conversion. yield calculated by the previous described model are under these
values due to external and internal mass transfer limitations. Val-
3.2. Characteristic times analysis ues of equilibrium conversion and yield are above the kinetic ones,
which is logical.
Numerical values of characteristic times are given in Table 4 for
two particle diameters. A residence time of one hour has been 3.4. Concentration profiles in the reactor
chosen because it ensures a high triolein conversion.
External mass transfer and diffusion time are calculated using Previous study showed that the internal mass transfer limita-
the effective diffusion of triolein. For a large value of particle diam- tions decrease drastically the reactor performances compare to
eter, the reaction is limited by internal mass diffusion so that the pure thermodynamically conditions. To understand the behavior
catalyst is working in diffusional regime. When catalyst particle of the system, concentration profiles and reaction rates are calcu-
size is decreased, diffusion time decreases indicating a faster phe- lated in the liquid phase and inside the catalyst particle using the
nomenon. This diffusion time remains higher (100 s) than the inlet finite volume model that has been described previously. Firstly,
reaction time of reaction (R1) (40 s). one laboratory reactor with a length of 0.24 m is simulated. The
catalyst particle diameter is fixed and equal to 2.5 mm. The meth-
3.3. Influence of inlet methanol to triolein ratio anol to triolein molar ratio is equal to 26 and the global residence
time is equal to 3600 s, a value sufficiently high compared to the
The methanol to triolein global molar ratio is defined by the fol- inlet reaction time of reaction (R1). These conditions guarantee a
lowing expression: reasonable conversion level. Triolein and methanol concentrations
F eMeOH at reactor inlet are respectively equal to 373 mol/m3 and
R¼ ð39Þ 10090 mol/m3. There is no other component at inlet. Conversion
F eT
is achieved at 175 °C under a pressure of 50 bar. The number of
As described in literature, the methanol to triolein ratio has a mesh steps is equal to 130 in the catalyst particle, 100 in the lon-
strong influence on triolein conversion and ester yield. The defini- gitudinal liquid phase and 1000 for time steps. Figs. 6 and 7 depict
tion of triolein conversion and ester yield are given by: respectively the concentrations profiles of triglyceride, diglyceride,

0.9

0.8

0.7

0.6
E
X ,Y
T

0.5

0.4

0.3

0.2

0.1
0 5 10 15 20 25 30 35 40
R

Fig. 5. Triolein conversion and ester yield as a function of the inlet molar ratio (+ Equilibrium conversion,  Equilibrium yield, ——— Kinetic conversion, – – Kinetic yield)
(dp = 2.5 mm, s = 3600 s).

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx 9

Tryglyceride concentration Diglyceride concentration

220
120
0.2 200 0.2
180
100
160
0.15 0.15
140 80
z (m)

z (m)
120
0.1 0.1
100 60

80
0.05 60 0.05 40

40
20
0 20 0
6 5 4 3 2 1 0 6 5 4 3 2 1 0
3 −8 3 −8
Particle volume (m ) x 10 Particle volume (m ) x 10

Monoglyceride concentration Glycerol concentration

240
70
220
0.2 65 0.2
200
60
180
0.15 55 0.15
160
z (m)

z (m)
50
140
45 120
0.1 0.1
40 100

35 80
0.05 0.05
30 60

25 40
0 0
6 5 4 3 2 1 0 6 5 4 3 2 1 0
Particle volume (m3) x 10
−8
Particle volume (m3) −8
x 10

Fig. 6. Single reactor, concentration profiles (mol/L) (R = 26, s = 3600 s).

monoglyceride and glycerol and the rates of reactions (R1)–(R3). where glycerol as a small molecule diffuses about 5 times faster
Ordinate represents the reactor length, and abscissa the volume than triolein.
of catalyst particle: at left side, the catalyst surface is depicted Monoglyceride and diglyceride concentrations maps present
ðr ¼ Rp ¼ 1:25 mmÞ, at right side, indication 0 corresponds to the maximal concentrations regions due to complex coupling of equi-
center of the particle (r = 0). librium consecutive reactions and internal mass diffusion in the
Results show that the concentration of triolein decrease catalyst particle.
strongly at the beginning of the reactor (iso-concentration curves Rates of reactions are presented in Fig. 7. Reaction (R1) is the
are very closed together in the first part of the reactor and more fastest reaction taking place in the first half of the reactor with a
separated in the second part). This means that conversion of trio- small fraction of catalyst. Thermodynamical equilibrium is also
lein is mainly realized in the first part of the reactor and with a quickly reached. Reactions (R2) and (R3) are slower and take place
low fraction of catalyst. in a larger part of the reactor. The whole catalyst is used.
The concentration of diglyceride increases quickly at the begin-
ning of the reactor as it is produced by reaction (R1) and then de- 4. Development of a strategy for conversion improvement
creases by reaction (R2). The same behavior is observed for the
concentration profile of monoglyceride but respectively with reac- The aim of this study is to determine if the global performance
tions (R2) and (R3). of the system can be improved with the use of an inter stage reac-
The concentration front corresponding to the maximal concen- tant addition. The case of methanol addition will be especially
tration of monoglyceride is slightly shifted forward in the reactor studied because methanol is introduced with a large excess.
with respect to diglyceride which confirms that reactions are con- The three reactions have a quite different rate (reaction (R1) is
secutive. A larger fraction of catalyst is used to realize the conver- fast whereas reactions (R2) and (R3) are slower). It is also difficult
sion of diglyceride and monoglyceride. to optimize the process with only one reactor. The addition of a
Concentration gradients in the catalyst for triolein, diglyceride second reactor could be useful in order to convert the main part
and monoglyceride are orthogonal to the main external fluid of triolein in the first one and to convert diglyceride and monoglyc-
flow and show that diffusion of these components in the catalyst eride in the second one. Methanol should be optimally injected be-
is slow. This is not the case of glycerol which presents collinear tween the first and the second reactor. It could also be possible to
profile of concentration with the fluid flow and also a better increase the temperature especially in the second reactor to en-
diffusion in the catalyst. These results are in agreement with hance rates of reactions (R2) and (R3), which are slow reactions,
the values of the diffusion coefficient presented in Table 1, but it has not been considered in this paper.

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
10 J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx

Reaction 1 x 10
−3 Reaction 2 x 10
−4

4
3
0.2 3.5 0.2

2.5
3
0.15 0.15
2.5 2
z (m)

z (m)
2
0.1 0.1 1.5
1.5
1
0.05 1 0.05

0.5 0.5
0 0
6 5 4 3 2 1 0 6 5 4 3 2 1 0
3 −8
3 −8
Particle volume (m ) x 10 Particle volume (m ) x 10

Reaction 3 x 10
−5

11
0.2
10

0.15 9
z (m)

0.1
7

6
0.05
5

4
0
6 5 4 3 2 1 0
Particle volume (m3) x 10
−8

Fig. 7. Single reactor, reaction rate (mol/kg/s) (R = 26, s = 3600 s).

Two laboratory reactors having each a length of 0.12 m have Q 1 ¼ Q Oil þ Q MeOH;1 ð43Þ
been considered. The same conditions as those presented in the
previous paragraph (particle size, methanol to triolein inlet molar Q 2 ¼ Q 1 þ Q MeOH;2 ¼ Q Oil þ Q MeOH ð44Þ
ratio, global residence time) are used and the setup of the reactors
QOil corresponds to the volumic flow rate of oil and QMeOH,2 rep-
is shown in Fig. 8.
resents the volumic flow rate of methanol in the second reactor. To
The study requires the definition of a new parameter, the volu-
find the optimal conditions corresponding to the maximal conver-
mic feed ratio a defined as follows:
sion rate, triolein conversion is plotted with respect to the inlet
Q MeOH;1 methanol to triolein molar ratio and the volumic feed ratio in
a¼ ð42Þ Fig. 9.
Q MeOH
Staging has an impact on the conversion rate of triolein (as well
QMeOH,1 represents the volumic flow rate of methanol in the first as on the yield) as pictured in Fig. 9. A relatively high conversion
reactor, divided by the total volumic flow rate of methanol QMeOH. can be reached at a high methanol to triolein molar ratio. This va-
A feed ratio equal to 1 means that the whole methanol is intro- lue can be slightly decreased when methanol is optimally intro-
duced in the first reactor. When the feed ratio is decreased to duced in each reactor.
0.1, it means that only 10% of the methanol flow rate is injected These curves show that the use of a molar ratio larger than
in the first reactor. about 25 has not a large impact on conversion: for a = 1, a conver-
Expressions of the flowing volume through reactor 1 and reac- sion of 0.87 is reached for R = 19 whereas a conversion of 0.90 is
tor 2 are respectively given by: reached for R = 26.
Theoretically, the followed path used to reach the thermody-
namic equilibrium has no influence on conversion. This means that
the staging will have no influence on triolein conversion. This is ob-
served in Fig. 9 for low inlet methanol to triolein global molar ratio
R: whatever the value of feed ratio a, the conversion remains con-
stant for a given inlet molar ratio.
For a high inlet methanol to triolein molar ratio R, the behavior
is not the same, triolein conversion depends on the value of the
feed ratio a. When the global inlet molar ratio is fixed, an increase
of the feed ratio implies an increase of conversion. This phenome-
Fig. 8. Reactors setup. non is observed for low values of the feed ratio: the methanol to

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx 11

1 0.9

0.8380.838 0.87 0.87


0.9
0.805
0.805 0.903 0.903

0.8
0.773
0.773
0.8

0.74
0.74

0.7 0.7

0.35
0.35 0.708
0.708

0.6
0.415
0.415 0.675
0.675 0.6
α

0.5
0.48
0.48
0.643
0.643

0.5
0.4
0.545
0.545

0.61
0.61
0.3
0.4

0.2

0.3
0.1
5 10 15 20 25 30 35
R

Fig. 9. Triolein conversion as a function of a and the entry molar ratio (sg = 3600 s, dp = 2.5 mm).

triolein molar ratio at the inlet of the first reactor is very low (de-
Table 5 spite the fact that the global ratio is high). The rate of reactions are
Methanol to triolein molar ratio and catalyst efficiency at inlet of each reactor
also slow (due to a low concentration of methanol) that implies a
(sg = 3600 s, R = 36, dp = 2.5 mm).
poor conversion of triolein, and consequently of diglyceride and
a 0.1 0.8 1 monoglyceride in the first reactor. Moreover, a large part of the
Methanol to triolein molar Inlet of 1st reactor 3.6 28.8 36 methanol is also used to convert triolein by the first reaction which
ratio Inlet of 2nd reactor 73.9 185.3 178.8 is the fastest one. A high quantity of methanol is introduced only at
Catalyst efficiency Inlet of 1st reactor 0.28 0.15 0.15
the inlet of the second reactor, implying dilution, leading to low
related to triolein gT Inlet of 2nd reactor 0.15 0.19 0.19
rates of reaction in reactor 2 too.

1
0.7

0.9
0.65

0.8
0.6
0.613
0.613 0.6410.641

0.585
0.585 0.668 0.668 0.55
0.7
0.558
0.558 0.696 0.696

0.5
0.6
0.53
0.53
α

0.502
0.502 0.723 0.723 0.45
0.5
0.475
0.475
0.4

0.4
0.419
0.419
0.35
0.364
0.364

0.3
0.309
0.309
0.3
0.254
0.254

0.2
0.198 0.25

0.1 0.2
5 10 15 20 25 30 35
R

Fig. 10. Triolein conversion as a function of a and the entry molar ratio (sg = 3600 s, dp = 5 mm).

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
12 J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx

Table 6 also very low indicating strong mass transfer limitations in


Methanol to triolein molar ratio and catalyst efficiency at inlet of each reactor catalyst. Indeed, the catalyst efficiency at the inlet of the first reac-
(sg = 3600 s, R = 36, dp = 5 mm).
tor decreases from 0.09 to 0.07 (representing a reduction of 22%)
a 0.1 0.5 1 when a increases from 0.5 to 1. The values of catalyst efficiencies
Methanol to triolein Inlet of 1st reactor 3.6 18 36 and methanol to triolein molar ratio are given in Table 6. The
molar ratio Inlet of 2nd reactor 60.4 76.8 69.3 staging has also an interest only when mass transfer limitations
Catalyst efficiency Inlet of 1st reactor 0.15 0.09 0.07 occur.
related to triolein gT Inlet of 2nd reactor 0.08 0.08 0.08
In order to understand and to describe the impact of the var-
iation of the feed ratio a, concentrations and reaction rates pro-
files are plotted in the catalyst and the fluid phase in Figs. 11
The corresponding values of methanol to triolein molar ratio at and 12 for a = 0.1. This value of the feed ratio enables to obtain
the inlet of each reactor and catalyst efficiencies (calculated by Eq. different curves as those previously depicted. The discontinuity
(14)) are given in Table 5. at a length of 0.12 m corresponds to the separation between
The same calculations have been performed for a particle diam- reactors 1 and 2. In the first reactor, the concentration gradients
eter of 5 mm. The conversion rate of triolein is plotted with respect of triolein, diglyceride and monoglyceride are globally collinear
to the methanol to triolein global molar ratio R and the feed ratio a to the main external fluid flow. This show that a small value
in Fig. 10. As expected, an increase of the particle diameter implies of a guarantee a good diffusion in the catalyst in the first reac-
a decrease of conversion due to higher mass transfer limitations. tor. This was not observed for a = 1 (see Figs. 6 and 7). In the
Moreover it can be shown that staging has a higher interest for second reactor, the main part of the methanol is added leading
5 mm than for 2.5 mm because the optimal value of conversion to a dilution of the other reactants and then to slow diffusion
is not obtained for a = 1. Indeed, on the iso-curve corresponding in the catalyst with respect to the rate of chemical reaction.
to a conversion value of 0.723, the optimal feed ratio does not The concentration gradients are different. Although a poor global
correspond to a = 1 but to a = 0.53. This result is very interesting conversion of triolein, the staging with a small value of the feed
because staging enables here to obtain a higher conversion rate ratio has also a positive impact from diffusion phenomena point
without using a too large excess of methanol. In fact, at high meth- of view in the first reactor.
anol to triolein molar ratio, the triolein concentration is very low, Rates of reactions are presented in Fig. 12. For reaction (R1),
especially in the first reactor when a tends towards 1. The catalyst thermodynamical equilibrium is reached. In the first reactor, rates
efficiency (calculated from Thiele modulus related to triolein) is of reactions (R2) and (R3) are quite slow because the methanol

Tryglyceride concentration Diglyceride concentration


600 240

220
0.2 500 0.2
200

180

0.15 400 0.15 160


z (m)

z (m)

140
300 120
0.1 0.1
100
200
80
0.05 0.05
60
100
40
0 0
6 5 4 3 2 1 0 6 5 4 3 2 1 0
Particle volume (m3) x 10
−8
Particle volume (m3) −8
x 10

Monoglyceride concentration Glycerol concentration

120
180
0.2 110 0.2
160
100
140
90
0.15 0.15
120
80
z (m)

z (m)

70 100
0.1 0.1
60 80

50 60
0.05 0.05
40 40
30
20
0 0
6 5 4 3 2 1 0 6 5 4 3 2 1 0
Particle volume (m3) x 10
−8
Particle volume (m3) −8
x 10

Fig. 11. Two reactors, concentration profiles (mol/L) (R = 26, sg = 3600s, a = 0.1).

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx 13

Reaction 1 x 10
−3 Reaction 2 x 10
−4

2
2.5
0.2 1.8 0.2

1.6
2
0.15 1.4 0.15
z (m)

z (m)
1.2
1.5
0.1 1 0.1
0.8
1
0.6
0.05 0.05
0.4

0.2 0.5
0 0
6 5 4 3 2 1 0 6 5 4 3 2 1 0
Particle volume (m3) −8
x 10 Particle volume (m3) −8
x 10

Reaction 3 x 10
−5

12

0.2 11
10
9
0.15
8
z (m)

7
0.1 6
5
4
0.05
3
2
0
6 5 4 3 2 1 0
3 −8
Particle volume (m ) x 10

Fig. 12. Two reactors, reaction rate (mol/kg/s) (R = 26, sg = 3600 s, a = 0.1).

concentration is low. The rates of reaction become logically higher strategies could be more interesting for improving the rate of these
in the second reactor as methanol is added. reactions. For instance a two steps reactor with increasing temper-
ature strategy could be promising.
In this paper, the addition of methanol at inter stage has been
5. Conclusions
studied considering a two steps reactor, motivated by the different
rates of ester formation reactions. The staging presents a great
Transesterification of vegetable oils, consisting in three com-
interest especially when strong mass transfer limitations take
bined series and parallel reversible reactions, require a large excess
place. In these conditions, the global quantity of methanol at the
of methanol to shift chemical equilibriums. In a continuous heter-
system inlet can be lowered with respect to a reactor without stag-
ogeneous catalytic reactor, mass transfer limitations modify locally
ing having the same characteristics (mass of catalyst, geometry).
the rate of equilibrium reaction which further increases the com-
plexity of the process. To improve the understanding of the com-
plex behavior of the coupled phenomena taking place, a two- Acknowledgments
dimensional numerical model of the reactor has been developed.
The model has been first used to determine the intrinsic chemical The authors wish to acknowledge CNRS and IFP Energies Nouv-
rate of the reactions and it has been shown the good agreement be- elles for its financial support.
tween the calculated results and the experimental ones.
The simulations have shown that the use of a high molar ratio of References
methanol to triolein (about 25) is necessary to shift equilibrium
such as the triolein conversion reaches 80%. Indeed, the calculated [1] G. Arzamendi, I. Campo, E. Arguinarena, M. Sanchez, M. Montes, L.M. Gandia,
Synthesis of biodiesel with heterogeneous NaOH/alumina catalysts:
internal concentration profiles show that the internal diffusion of comparison with homogeneous NaOH, Chem. Eng. J. 134 (2007) 123–130.
triglyceride, diglyceride and monoglyceride is the limiting phe- [2] N. Barakos, S. Pasias, N. Papayannakos, Transesterification of triglycerides in
nomena in the catalyst, which was confirmed by a characteristic high and low quality oil feeds over an HT2 hydrocalcite catalyst, Bioresource
Technol. 99 (2008) 5037–5042.
time analysis. The conversion of triolein proceeds much faster than
[3] L. Bournay, D. Casanave, B. Delfort, G. Hillion, J.A. Chodorge, New
the other reactions and thermodynamical equilibrium is reached heterogeneous process for biodiesel production: a way to improve the
for reaction 1 whereas conversions of diglyceride (reaction (R2)) quality and value of the crude glycerin produces by biodiesel plants, Catal.
and monoglyceride (reaction (R3)) are much lower. This may Today 106 (2005) 190–192.
[4] J.M. Commenge, L. Falk, J.P. Corriou, M. Matlosz, Analysis of microstructured
explain the reason of large excess of methanol to enhance reactor characteristics for process miniaturization and intensification, Chem.
these two slow secondary reactions. These results show that other Eng. Technol. 28 (2005) 446–458.

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106
14 J.F. Portha et al. / Chemical Engineering Journal xxx (2012) xxx–xxx

[5] M. Diasakou, A. Louloudi, N. Papayannakos, Kinetics of the non-catalytic [10] D. Kusdiana, S. Saka, Kinetics of transesterification in rapeseed oil to biodiesel
transesterification of soybean oil, Fuel 77 (1998) 1297–1302. fuel as treated in supercritical methanol, Fuel 80 (2001) 693–698.
[6] T.F. Dossin, M.F. Reyniers, G.B. Marin, Kinetics of heterogeneously MgO- [11] S.V. Patankar, Numerical Heat Transfer and Heat Flow, Hemisphere Publishing
catalyzed transesterification, Appl. Catal. B 61 (2006) 35–45. Corporation, New York, 1980.
[7] T.F. Dossin, M.F. Reyniers, R.J. Berger, G.B. Marin, Simulation of [12] V. Pugnet, S. Maury, V. Coupard, A. Dandeu, A.A. Quoineaud, J.L. Bonneau, D.
heterogeneously MgO-catalyzed transesterification for fine-chemical and Tichit, Stability, activity and selectivity study of a zinc aluminate
Biodiesel industrial production, Appl. Catal. B 67 (2006) 136–148. heterogeneous catalyst for transesterification of vegetable oil in batch
[8] L. Falk, J.M. Commenge, Characteristics time analysis for process reactor, Appl. Catal. A: General 374 (2010) 71–78.
intensification, in: CHISA 2006–17th International Congress of Chemical and [13] D. Schweich (coordinator), Génie de la Réaction Chimique, Lavoisier Tec et Doc,
Process Engineering. Paris, 2001.
[9] H.J. Kim, B.S. Kang, M.J. Kim, Y.M. Park, D.K. Kim, J.S. Lee, K.Y. Lee, [14] J. Villermaux, Génie de la Réaction Chimique, Conception et Fonctionnement
Transesterification of vegetable oil to biodiesel using heterogeneous base des Réacteurs, Lavoisier Tec et Doc, Paris, 1993.
catalyst, Catal. Today 93–95 (2004) 315–320.

Please cite this article in press as: J.F. Portha et al., Simulation and kinetic study of transesterification of triolein to biodiesel using modular reactors, Chem.
Eng. J. (2012), http://dx.doi.org/10.1016/j.cej.2012.06.106

Vous aimerez peut-être aussi