Академический Документы
Профессиональный Документы
Культура Документы
www.elsevier.com/locate/ces
Received 6 February 2006; received in revised form 7 October 2006; accepted 17 October 2006
Available online 28 October 2006
Abstract
The synthesis of dimethylacetal using acetaldehyde and methanol as raw material in the presence of Smopex 101 fibres as catalyst and
adsorbent in batch reactor and in a fixed bed adsorptive reactor, respectively, was studied. In the batch reactor the determination of thermodynamic
and reaction kinetics data for acetalization reaction was presented. A kinetic model based on a pseudo-homogeneous rate expression using
activities was proposed to describe the experimental kinetic results. The dynamic binary adsorption experiments were carried out in the absence
of reaction at 293.15 K in a laboratory scale column. The experimental results of the adsorption of binary non-reactive mixtures are reported
and used to obtain multi-component adsorption equilibrium isotherms of Langmuir type. The mathematical model was proposed to describe the
adsorptive reactor dynamic behaviour. The experimental results obtained for the reaction and regeneration experiments were compared with the
model proposed. Model equations were solved by orthogonal collocation on finite elements (OCFE) implemented by the PDECOL package,
using the measured model parameters and was validated for both reaction and regeneration steps.
䉷 2006 Elsevier Ltd. All rights reserved.
Keywords: Acetaldehyde dimethylacetal; Adsorption; Batch catalytic reactor; Fixed bed reactor; Kinetics; Smopex 101 fibres; Thermodynamics
0.40
2103.3
ln Keq = − 4.0595. (1) 283 K
T (K) 0.30
293 K
0.20 303 K
3.2. Reaction kinetics
0.10 313 K
1.00 Table 3
The kinetic parameter as a function of temperature
0.90
Temperature (K) kC (mol/g min)
Acetaldehyde Conversion
0.80
0.70 293.15 1.67
303.15 2.92
0.60
313.15 5.99
0.50
0.40 1.5
0.30 2.0
analogy with the homogeneous catalysis, the simple mechanism
0.20 3.0
of the above reaction for the synthesis of dimethylacetal for
0.10 4.0
homogeneous is given in three steps (Morrison and Boy, 1983;
0.00 Chopade and Sharma, 1996); the reaction between acetaldehyde
0 50 100 150 200 250 300 and methanol leads to the formation of intermediate hemiacetal
Time (min) which is not stable. The hemiacetal is then reacted with the
second methanol molecule to form the final product acetal. The
Fig. 3. Effect of initial molar ratio of reactants (methanol/acetaldehyde) on
the acetaldehyde conversion as function of time: T = 293.15 K, P = 6 atm,
various steps that take place during this reaction are shown
V = 600 mL, wcat = 0.5 g. below.
Step 1: Surface reaction of methanol (A) and acetaldehyde
(B) to give hemiacetal (I1 ):
0.80
A + B ←→ I1 .
0.70
Aldehyde Conversion
0.40 H+
0.042 % w/V I1 ←→ I2 + D (rate controlling step).
0.30
0.083 % w/V
0.20 0.125 % w/V
Step 3: Surface reaction of formation of acetal (C):
0.167 % w/V
0.10 H+
I2 + A ←→ C.
0.00
0 50 100 150 200 250
Considering this pseudo-homogeneous mechanism, the reac-
Time (min)
tion rate equation is then given by
Fig. 4. Effect of catalyst loading on acetaldehyde conversion as function of
time: T = 293.15 K, P = 6 atm, V = 600 mL, rA/B = 2.0. aC a D
R = kC aA a B − . (2)
Keq aA
increase in the catalyst loading; however, after a certain value, This rate expression is in accordance with the Langmuir–
further increases in catalyst loading leads to a marginal increase Hinshelwood model when the adsorption constants for all
in acetaldehyde conversion. Obviously, the equilibrium com- components are negligible.
position will be the same for all cases, just the time needed to The kinetic constant (kC ) at each temperature was obtained
achieve it will be higher for low catalyst loadings. from the experimental data by optimization. The values of the
kinetic parameter kC are presented in Table 3. The temperature
3.3. Kinetic modeling dependence of the kinetic constant was fitted with the Arrhenius
equation (Fig. 5), where the activation energy obtained was of
The kinetic model based on a Langmuir–Hinshelwood rate 48.67 kJ/mol. The Arrhenius equation fitted to the experimental
expression for acetalization reactions was already derived in kinetic parameters is given below:
previous works (Silva and Rodrigues, 2001, 2006). However,
in this case the methodology used is different because it is −5853.8
kC = 7.59 × 10 exp8
(mol/g min),
supposed that the reaction takes place on the active sites on the T (K)
surface of the catalyst due to the absence of the pores in the r 2 = 0.9914. (3)
Smopex 101 catalyst. Since the surface area of Smopex fibres is
much lower than the specific area of the available heterogeneous Fig. 6 shows the comparison between experimental and sim-
macroporous catalysts (such as Amberlyst 15), the adsorption ulated results for acetaldehyde, methanol and products (acetal
of the species at the surface can be negligible. Therefore, by and water) at 293.15 K. The average standard error between
G.K. Gandi et al. / Chemical Engineering Science 62 (2007) 907 – 918 911
the experimental and simulated molar fractions of all species 4. Fixed adsorptive reactor
is 2.0%. The kinetic model is in agreement with experimental
results for the range of initial molar ratio of reactants from 1.5 4.1. Experimental set-up
to 4.0 and at the temperature range 293.15–313.15 K.
The laboratory-scale experimental set-up (Gandi et al.,
2006) and the characteristics of the column bed are given in
Table 4. The bed was prepared by introducing glass beads in
2
the jacketed glass column together with the Smopex 101 fibres
in order to avoid building of pressure in the column due to the
1.5 compression of the bed packed only with Smopex 101 fibres.
A known amount of Smopex 101 slurry was prepared with the
Ln kc
1
Table 4
Characteristics of the fixed bed reactor
0.5
Solid weight (Smopex 101 fibres) 3.4 × 10−3 kg
Length of bed (L) 9.5 × 10−2 m
0 Internal diameter (Di ) 2.6 × 10−2 m
0.0031 0.0032 0.0033 0.0034 0.0035 Bulk density of smopex fibres (b ) 410 kg/m3
1/T Apparent density of Smopex fibres (f ) 567 kg/m3
Mass of glass beads 0.076 kg
Diameter of glass bead 0.003 m
Fig. 5. Representation of the experimental values of ln[k(mol/g min)] as
Volume occupied by glass beads 31.0 × 10−6 m3
function of 1/[T (K)] and respective linear fitting.
a b
9 10
Methanol 9 Methanol
8 T=293.15, T=293.15,
rA/B = 2.0 Products
7 rA/B = 1.50 Products 8
Acetaldehyde
Number of Moles
Number of Moles
Aldehyde 7
6
True Kinetics True Kinetics
6
5
5
4
4
3
3
2 2
1 1
0 0
0 50 100 150 200 250 0 100 200 300 400
Time (min) Time (min)
c d
12 12
Methanol Methanol
T=293.15, T=293.15,
10 rA/B = 3.0 Products 10 rA/B = 4.0 Products
Aldehyde Aldehyde
Number of Moles
Number of Moles
8 8
True Kinetics True Kinetics
6 6
4 4
2 2
0 0
0 100 200 300 400 0 50 100 150 200 250 300
Time (min) Time (min)
Fig. 6. Comparison between experimental and true kinetic model curves at T = 293.15 K, P = 6 atm, V = 600 mL, wcat = 0.50 g. (a) rA/B = 1.50; (b) rA/B = 2.0;
(c) rA/B = 3.0; (d) rA/B = 4.0.
912 G.K. Gandi et al. / Chemical Engineering Science 62 (2007) 907 – 918
8 lion or moles per unit volume or mass per unit volume. The
54 3 2 concentration of adsorbent on the solid is given as moles or
7 1 1 Q= 2 mL/min
mass adsorbed per unit mass or unit volume of original adsor-
2 Q= 3 mL/min
bent. Some authors estimated the mono- and multi-component
Detector Response (mV)
6 3 Q= 4 mL/min
4 Q= 5 mL/min isotherms based on the swelling ratio values of the polymeric
5 5 Q= 6 mL/min adsorbent. The swelling ratio values for Smopex 101 adsor-
bent at 20 ◦ C are 1.38 for water, 1.25 for ethanol and 1.18
4
for propanol (Lilja, 2005). More details about the swelling be-
3 haviour of Smopex 101 fibres for mono and binary systems are
also presented in Lilja (2005). In this work, the methodology is
2 based on experimental results of dynamic adsorption of binary
non-reactive mixtures to obtain binary adsorption equilibrium
1
data. Therefore, since no adsorption data for single component
0 were measured, the multi-component adsorption equilibrium
0 4 8 12 16 20 related to the fibre surface concentration was described by the
Time (min) extended Langmuir model:
The above mass balance equations are associated with the initial where Shp = ke dp /Dm and Rep = dp u/
are, respectively,
and Danckwerts boundary conditions: the Sherwood and Reynolds numbers relative to the particle,
and Sc =
/Dm is the Schmidt number.
at t = 0,Ci = Cs,i = Ci,0 , (7) The diffusivities in multi-component liquid mixture were es-
jCi timated by the modified Wilke–Chang equation, proposed by
z = 0, uC i − εDax = uC i,F , (8a)
jz z=0 Perkins and Geankoplis (Reid et al., 1988):
jCi (M)1/2 T
z = L, = 0, (8b) o
DAm = 7.4 × 10−8 , (18)
jz z=L
Vml,A
0.6
n
ficial velocity, Kf is the external film mass transfer coefficient, M = x j j M j , (19)
Dax is the axial dispersion coefficient, t is the time variable, j =1
z is the axial coordinate, i is the stoichiometric coefficient of j =A
component i, f represents the apparent density of fibres, R is o is the diffusion coefficient for a dilute solute A into
where DAm
the rate given by Eq. (2), the subscripts F and 0 refer to the feed
a mixture of n components, T is the temperature, Vml,A is the
and initial states, respectively, and subscript f represents fibres.
molar volume of the solute A, xj is the molar fraction for com-
Introducing the parameter space time = εL/u and the di-
ponent j, j is the association factor for component j, while is
mensionless variables for space = z/L nand time = t/, equal to 2.6 for water, 1.9 for methanol and 1.0 for unassoci-
C̃
n = C/C F,T , q̃ = q/q F,T where C F,T = i=1 CF,i and qF,T = ated components. The mixture viscosity
was predicted by the
i=1 qi (CF,1 , CF,2 , . . . , CF,n ), and the parameters, generalized corresponding state method proposed by Teja and
uL Rice (1981).
Pe = (Peclet number), (9)
εDax The system of partial differential equations represented by
Eqs. (13)–(16) are solved by the method of lines (MOL) using
kc
Da = (Damköhler number), (10) orthogonal collocation in finite elements to discretize the axial
qF,T coordinate, with B-splines as base functions through numerical
1−ε 2 package PDECOL. Fifty subintervals for spatial discretization
Nf,i = f Kf,i along the z-axis were used, with two internal collocation points
ε rp
(number of film mass transfer units), (11) in each subinterval, resulting in 200/400 time-dependent ordi-
nary differential equations for adsorption/reaction simulations.
(1 − ε)f f qF,T For all simulations a tolerance value EPS equal to 10−7 was
=
εCF,T fixed.
(mass capacity factor at feed conditions). (12)
The above model Eqs. (5) and (6), respectively, become as 4.4. Experimental results and discussions
follows:
4.4.1. Binary adsorption experiments
jC̃i jC̃i 1 j j2 xi The breakthrough curves of methanol, acetal (DME) and wa-
+ + Nf,i (C̃i − C̃s,i ) = C̃T 2 , (13)
j j P e j j ter were measured in the absence of reaction. The adsorbent
was saturated with a certain component A, and then the feed
jq̃is with pure component B was fed to the adsorbent on stepwise
Nf,i (C̃i − C̃s,i ) = − i
j concentration of the feed solution. The adsorption parameters
× Da(as,A as,B − as,C as,D /(Keq as,A )), were optimized by minimizing the difference between the ex-
(14) perimental and simulated stoichiometric times of breakthrough
curves. The stoichiometric time can be determined from the
with initial and boundary conditions
mass balance over the adsorbent bed in the column as shown
= 0, Ci = Cs,i = C0,i , (15) below:
1 jCi L ∗ q
= 0, Ci − = Ci,F , (16a) tf = ε + f , (20)
P e j =0 u C
jCi where the parameter ∗f is defined as
= 1, = 0. (16b)
j =1 mf
∗f = (21)
The external mass transfer coefficient was estimated by the Vb
Wilson and Geankoplis correlation (Ruthven, 1984):
and mf is the mass of fibres and Vb is the bed volume in the
1.09 column. The adsorption amount q is expressed in number of
Shp = (Rep Sc)0.33 , 0.0015 < Rep < 55, (17)
ε moles adsorbed per mass of fibres (mol/kgfibre ).
914 G.K. Gandi et al. / Chemical Engineering Science 62 (2007) 907 – 918
30 60
Methanol
25 50
Theoretical 40
20
30 Methanol
15
Water
20
10 Theoretical
10
5
0
0 0.0 5.0 10.0 15.0 20.0 25.0 30.0
0 50 100 150 Time (min)
Time (min)
60
30
50 Methanol
Theoretical
20
30
Methanol
15
Acetal 20
10 Theoretical
10
5
0
0 50 100 150 200 250
0
0.0 10.0 20.0 30.0 40.0 Time (min)
Time (min)
Fig. 9. Breakthrough experiments, outlet concentration of methanol and wa-
ter as a function of time. (a) Water displacing methanol; Q = 4.0 mL/ min,
Fig. 8. Breakthrough experiments, outlet concentration of acetal and methanol
bottom-up flow direction; (b) methanol displacing water; Q = 4.0 mL/ min,
as a function of time. (a) Acetal displacing methanol; Q = 4.0 mL/ min,
top-down flow direction. Experimental data: () methanol; ( ) water. The-
bottom-up flow direction; (b) methanol displacing acetal; Q = 4.0 mL/ min,
oretical results: (—).
top-down flow direction. Experimental data: () methanol; () acetal. The-
oretical results: (—).
The apparent density of the fibres can be calculated by of acetal, Fig. 8(a), the errors between the experimental and
∗f theoretical amounts of acetal adsorbed and methanol displaced
mf
f = = , (22) are 3.7% and 6.4%, respectively. Comparing the experimental
f (1 − ε) Vf amount of methanol adsorbed in the experiment of Fig. 8(b)
where with the experimental amount of methanol desorbed in the
experiment of Fig. 8(a), the error obtained is 4.9%. For acetal,
q q(CF ) − q(C0 )
= . (23) the error between the amount adsorbed, Fig. 8(a), and the
C CF − C 0 amount eluted, Fig. 8(b), is 3.8%.
Experimental stoichiometric times of the breakthrough Similarly, experiments were conducted for the combination
curves were calculated from the plot of outlet concentration of methanol and water breakthroughs and the breakthrough
of column and time of outlet samples analyzed. In the break- plots for methanol and water are shown in Figs. 9(a,b);
through experiments, where component A displaces component the experimental results are in agreement with the model
B, the total amount of A necessary to saturate the column is predictions.
given by the product of volumetric flow rate and the area over Using the experimental data and theoretical parameters, ad-
the breakthrough curve limited by the feed concentration. The sorption of methanol, acetal (DME) and water on the adsorbent
product between the flow rate and the area under the effluent (Smopex 101) were calculated at different steps of feed solu-
curve gives the total amount of component B that was ini- tions so as to determine the adsorption isotherms. Adsorption
tially saturating the column. In Figs. 8(a, b), the concentration parameters for methanol, acetal and water were calculated with
graphs of methanol and acetal are shown. In the breakthrough the data obtained from the stepwise and binary experiments
G.K. Gandi et al. / Chemical Engineering Science 62 (2007) 907 – 918 915
Table 6
Adsorption isotherms 30
Component Q 3
(mol/mreal K (m3 /mol)
solid )
25
0.93 × 10−3 Theoretical
30 6
20 2.0 min
4
4.0 min
15 3
6.0 min
10 2
8.0 min
8.0 min
5 1 2.0 min 4.0 min 6.0 min
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Dimensionless axial distance Dimensionless axial distance
5
6
4.5 8.0 min
4 5 8.0 min
3.5 2.0 min 4.0 min 6.0 min 2.0min 4.0 min 6.0min
4
Acetal Conc. (mol/L)
3
2.5 3
2
1.5 2
1
1
0.5
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Dimensionless axial distance Dimensionless axial distance
Fig. 11. Internal concentration profiles of each species in fluid phase inside the column, at different times, during the reaction experiment of Fig. 10(a): (a)
methanol, (b) acetaldehyde, (c) acetal and (d) water.
5. Conclusions Notation
Q adsorption capacity, mol/kg Chopade, S.P., Sharma, M.M., 1997. Reaction of ethanol and formaldehyde:
rp particle radius, m use of versatile cation-exchange resins as catalyst in batch reactors and
reactive distillation columns. Reactive and Functional Polymers 32, 53–64.
rA/B initial molar ratio of reactants
Doraiswamy, L.K., Sharma, M.M., 1987. Heterogeneous Reactions: Analysis,
R rate of reaction, mol/kg s Examples, and Reactor Design. Wiley-Interscience, New York.
t time, s Dupont, P., Védrine, J.C., Paumard, E., Hecquet, G., Lefebvre, F., 1995.
T temperature, K Heteropolyacids supported on activated carbon as catalysts for the
u superficial velocity, m/s esterification of acrylic acid by butanol. Applied Catalysis A 129,
217–227.
V solution volume, m3
Fedriani, A., Gonzalez-Velasco, J.R., Lopez-Fonseca, R., Gutierrez-Ortiz,
Vml molar volume of component i, m3 /mol M.A., 2000. Kinetics of 1,1-diethoxyethane synthesis catalysed by USY
wcat mass of catalyst, kg zeolites. CHISA 2000, A7.7 12.30 presentation.
x molar fraction Fredeslund, A., Gmehling, J., Rasmussen, P., 1977. Vapor–Liquid Equilibrium
X conversion of the limiting reactant Using UNIFAC. Amsterdam, Elsevier.
Gandi, G.K., Silva, V.M.T.M., Rodrigues, A.E., 2005. Process development
z axial coordinate, m
for the dimethylacetal synthesis: thermodynamics and reaction kinetics.
Industrial Engineering and Chemistry Research 44, 7287–7297.
Greek letters Gandi, G.K., Silva, V.M.T.M., Rodrigues, A.E., 2006. Synthesis of 1,1-
dimethoxyethane in fixed bed reactor. Industrial Engineering and Chemistry
activity coefficient Research 45, 2032–2039.
Gupta, B.B.G., 1987. Alkoxyalkylation of phenol. US Patent 4, 694,111.
ε porosity
Kariner, R.S., Krause, A.O.I., 2001. Kinetic model for the etherfication
Silva, V.M.T.M., Rodrigues, A.E., 2001. Synthesis of diethyl acetal: Teja, A., Rice, P., 1981. Generalized corresponding states method for
thermodynamic and kinetic studies. Chemical Engineering Science 56, the viscosities of liquid mixtures. Industrial and Engineering Chemistry
1255–1263. Fundamentals 20, 77.
Silva, V.M.T.M., Rodrigues, A.E., 2006. Kinetic studies in a batch reactor Yadav, G.D., Kishnan, M.S., 1998. An ecofriendly catalytic route for the
using ion-exchange resin catalysts for oxygenates production: the role of preparation of perfumery grade methyl anthranilate from anthranilic acid
mass transfer mechanisms. Chemical Engineering Science 61, 316–331. and methanol. Organic Process Research Development 2, 86–95.
da Silva-Machado, M., Cardoso, D., Pérez-Pariente, J., Sastre, E., 2000. Yadav, G.D., Pujari, A.A., 1999. Kinetics of acetalization of perfumery
Esterification of lauric acid with glycerol using modified zeolite beta as aldehydes with alkanols over solid acid catalysts. Canadian Journal of
catalyst. Studies in Surface Science and Catalysis 130D, 3417–3422. Chemical Engineering 77, 489–496.
Tanaka, K., Yoshikawa, R., Ying, C., Kita, H, Okamoto, K.I., 2002.
Application of zeolite T membrane to vapor-permeation-aided esterification
of lactic acid with ethanol. Chemical Engineering Science 57 (9),
1577–1584.