Glycerol valorisation as biofuels: Selection of a suitable solvent for an

innovative process for the synthesis of GEA

Rui P.V. Faria, Carla S.M. Pereira

, Viviana M.T.M. Silva, Jos M. Loureiro, Alrio E. Rodrigues

Laboratory of Separation and Reaction Engineering, Department of Chemical Engineering, Faculty of Engineering of University of Porto, 4200-465 Porto, Portugal
h i g h l i g h t s
Solvent selection for reactive
adsorptive processes: Synthesis of
glycerol acetal.
Acetalization of glycerol over
Amberlyst-15 in the presence of
dimethyl sulfoxide.
Kinetic model considering a LHHW
mechanism and diffusion inside the
catalyst.
g r a p h i c a l a b s t r a c t
a r t i c l e i n f o
Article history:
Received 30 March 2013
Received in revised form 6 August 2013
Accepted 12 August 2013
Available online 21 August 2013
Keywords:
Glycerol acetal
Acetalization
Amberlyst-15
Solvent selection
SMBR
a b s t r a c t
The study of the production of glycerol ethyl acetal, through the acetalization of glycerol with acetalde-
hyde in a simulated moving bed reactor requires the selection of a suitable solvent/desorbent for the pro-
cess. The methodology described in this paper, based on reaction, adsorption and environmental
variables, lead to the identication of dimethyl sulfoxide as the solvent that best tted the specications
of such reactiveadsorptive processes. Furthermore, a detailed study of the acetalization of glycerol in the
presence of this solvent and using Amberlyst-15 as catalyst, allowed the determination of its reaction
activation energy, 77.7 kJ mol
1
. The reaction rate was described by a LangmuirHinshelwoodHou-
genWatson model, and was obtained from batch experiments with a complete model considering dif-
fusion inside the catalyst particles.
2013 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
1. Introduction
For the past few years, biodiesel has been suggested as a reliable
renewable alternative for fossil fuels, mainly due to its reduced
toxicity and exhaust emissions. However, the intensication of
the production of these fatty acid esters [1] lead to an excess of
glycerol in the market, since it is a byproduct obtained in approx-
imately 10%, wt/wt, motivating the research and development of
several glycerol-based products [24].
Glycerol-acetals have shown interesting results as green oxy-
genated fuel additives [3,5,6]. Within this family of compounds,
glycerol ethyl acetal (GEA), has evidenced the potential to reduce
exhaust gases particle emissions [7,8] and control of fuel uid
properties [913]. However, it must be mentioned that GEA might
not be a suitable gasoline additive, due to the existence of solubil-
ity problems [14]. On the other hand, this problem was not veried
for diesel or biofuels [15,16].
Researchers have reported the acetalization of glycerol with
acetaldehyde [1724] and the transacetalization of glycerol with
1,1-diethoxyethane (DEE) [8,2528] as the two main reaction
pathways for the synthesis of GEA. In both cases an isomeric mix-
ture containing cis-5-hydroxy-2-methyl-1,3-dioxane, trans-5-hy-
droxy-2-methyl-1,3-dioxane, cis-4-hydroxymethyl-2- methyl-1,3-
dioxolane, and trans-4-hydroxymethyl-2-methyl-1,3- dioxolane
is obtained. However, the direct condensation of glycerol with
http://dx.doi.org/10.1016/j.cej.2013.08.035
1385-8947/ 2013 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

Corresponding author. Tel.: +351 22 508 1686; fax: +351 22 508 1674.
E-mail address: cpereir@fe.up.pt (C.S.M. Pereira).
Chemical Engineering Journal 233 (2013) 159167
Contents lists available at ScienceDirect
Chemical Engineering Journal
j our nal homepage: www. el sevi er . com/ l ocat e/ cej
acetaldehyde not only requires a less complex production scheme
but it also has a higher mass efciency and only makes use of
reactants obtained from biomass, presenting itself as a more inter-
esting route for the synthesis of GEA. Nonetheless, this route still
presents some drawbacks. On one hand, this is a thermodynami-
cally limited reaction. On the other hand, as the product is in-
tended to be applied as a fuel additive, the water formed as a
secondary product during the synthesis step must only be present
in residual amounts in the nal product. Therefore, the assay re-
quired for GEA might imply signicant separation costs.
The concept behind multifunctional reactors, which integrate
reaction and separation processes within a single unit, relies on
the equilibrium displacement promoted by the continuous re-
moval of the products from the reaction media, favouring products
formation and, therefore, increasing conversion. These hybrid tech-
nologies are able to improve selectivity towards a product and also
have the potential to reduce capital investment and energy con-
sumption. In what concerns acetals production, very satisfactory
results have been obtained with the simulated moving bed reactor
(SMBR) [29]. The production of 1,1-dimethoxyethane [30], DEE
[3132] and 1,1-dibuthoxyethane [33] was achieved by the direct
acetalization of methanol, ethanol and butanol with acetaldehyde,
respectively, in a SMBR where the chromatographic columns
where packed with Amberlyst-15 that acted both as catalyst and
adsorbent. Thus, the SMBR might be a suitable technology for the
production of GEA from glycerol and acetaldehyde, once it will
be able to shift the reaction thermodynamic equilibrium and,
simultaneously, remove the undesired water from the nal prod-
uct stream as veried for the previous acetalizations in this multi-
functional reactor.
In most SMBR applications reported in the open literature, one
of the reactants was used as solvent; however, in the GEA synthesis
by direct condensation of glycerol with acetaldehyde, this method-
ology cannot be applied. Glycerol, due to its high viscosity would
signicantly increase the pressure drop in the operation unit as
well as add a mass transfer resistance to the system. Plus, its high
boiling point would represent an increase in the nal product
purication costs. Acetaldehyde, on the other hand, in the presence
of an acid catalyst will undergo oligomerization reactions [34],
thus in order to decrease the possibility of side-products forma-
tion, acetaldehyde amount must be kept as low as possible. There-
fore, another compound must be selected as solvent and particular
attention was given to the analysis of parameters related to its
reactivity and physical properties with focus on adsorption. As
far as our knowledge goes, although some systematic solvent
screening methodologies have been published, none was speci-
cally focused on SMBR. They are either for generic organic
synthesis processes [3537], for simpler chromatographic systems
[3840] or mainly oriented for the identication of "green"
solvents [4143].
Therefore, the work herein reported, presents a rst approach
for a simple but expeditious solvent screening methodology for
reactiveadsorptive processes, such as the SMBR. A compilation
of signicant physicalchemical properties of a large list of the
most common solvents for chemical processes was performed
and the list was sequentially reduced based on theoretical param-
eters related with reactiveadsorptive processes requirements.
Miscibility, adsorption and reaction experimental tests were per-
formed with the solvents that presented the best results of the the-
oretical screening, allowing the selection of a suitable solvent for
the production of GEA in a SMBR.
Finally, a kinetic model was proposed for the direct condensa-
tion of glycerol with acetaldehyde in the presence of the selected
solvent. A parametric study comprising a wide range of operating
conditions was performed in order to determine the kinetic param-
eters for this reaction. The results were compared with the solvent
free reaction for the synthesis of GEA.
2. Experimental
2.1. Chemicals and catalyst
The chemicals used in the experimental tests were
acetaldehyde (>99%), glycerol (>99%), acetonitrile (>99%) and
Nomenclature
C
b,i
molar concentration of compound i in the bulk
(mol dm
3
)
C
p,i
molar concentration of compound i in the intraparticle
uid (mol dm
3
)
D
eff,i
compound i effective diffusion coefcient (cm
2
s
1
)
E
a
reaction activation energy (kJ mol
1
)
DG
0
standard reaction Gibbs free energy (kJ mol
1
)
DH
S,W
water enthalpy of adsorption (kJ mol
1
)
DH
S,DMSO
dimethyl sulfoxide enthalpy of adsorption (kJ mol
1
)
K
eq
reaction equilibrium constant ()
k
c
reaction kinetic constant (dm
6
mol
1
g
1
cat
s
1
)
k
c
0
Arrhenius pre-exponential factor for the reaction kinetic
constant (dm
6
mol
1
g
1
cat
s
1
)
K
S,W
water adsorption equilibrium constant (dm
3
mol
1
)
K
S;W
0
Vant Hoff pre-exponential factor for water adsorption
equilibrium constant (dm
3
mol
1
)
K
S,DMSO
dimethyl sulfoxide adsorption equilibrium constant
(dm
3
mol
1
)
K
S;DMSO
0
Vant Hoff pre-exponential factor for dimethyl sulfoxide
adsorption equilibrium constant (dm
3
mol
1
)
NP total number of measurements for a kinetic experiment
()
NE number of kinetic experiments performed ()
P pressure bar
r radial position (cm)
r
Acet/Gly
reactants initial molar ratio ()
q
ads
adsorbed molar amount (mol L
1
cat
)
R ideal gas constant (J mol
1
K
1
)
R
2
correlation coefcient ()
R
p
catalyst particle radius (lm)
R reaction rate based on the local intraparticle composi-
tion (mol g
1
cat
s
1
)
T temperature (K)
t time (min)
wt
cat
/wt catalyst weight percentage (%)
x
i
molar fraction of compound i ()
X
exp
ij
jth point for the conversion measured in experiment i (
)
X
mod
ij
jth point for the conversion in experiment predicted by
the model i ()
e
b
bulk porosity ()
e
p
catalyst particle porosity ()
m
i
compound i stoichiometric coefcient ()
q
p
catalyst particle solid density
Abbreviations
DEE 1,1-diethoxyethane
GEA glycerol ethyl acetal
SMBR simulated moving bed reactor
160 R.P.V. Faria et al. / Chemical Engineering Journal 233 (2013) 159167
n,n-dimethylformamide (>99%), purchased from SigmaAldrich,
and dimethyl sulfoxide (>99%) purchased from VWR International.
Rohm & Haas Amberlyst-15 ion exchange resin was used as cat-
alyst. As water adsorbs strongly on Amberlyst-15 and it can lead to
a decrease of the reaction rate, since it is a product of the reaction,
anhydrous resin must be used on adsorption and reaction tests.
Hence, the resin was washed several times with deionized water,
then washed with ethanol and dried at 363 K until the mass re-
mained constant, prior to its use.
2.2. Solvent selection methodology for reactiveadsorptive process
The solvent selection methodology developed comprises misci-
bility, reactivity, adsorption and kinetic tests with the solvents that
were selected after a rst theoretical screening.
2.2.1. Miscibility tests
Miscibility tests were performed by preparing several binary
mixtures containing the solvent and one of the reactants to which
water or GEA was added in low amounts until phase split or the
formation of a single phase system was observed (depending on
if the initial binary mixture presented one or two phases, respec-
tively). The tests were performed in 5 mL closed test-tubes kept
at 313 K by a water bath (LAUDA, LaudaKonigshofen, Germany).
The samples were initially stirred in a vortex mixer and then al-
lowed to stand to promote the phase separation.
2.2.2. Adsorption tests
Amberlyst-15 adsorption capacity towards water, GEA and the
solvents, was estimated through swelling experiments performed
in 10 mL graduated cylinders where a known amount of sorbent
(approximately 1.0 g) was put in contact with 5 mL of sorbate at
a constant temperature of 293 K (temperature was kept this low
in order to avoid evaporation) for a period of 48 h. The nal sorbent
volume was registered. The sorbent was ltered and weighted and
its adsorption capacity for a given compound was considered to be
an approximation of its mass difference between the beginning
and the end of the experiments. The procedure was performed at
least three times for each compound.
2.2.3. Reaction tests
Batch reactions were carried out in a 150 mL stainless steel ves-
sel, equipped with a sampling valve and temperature and pressure
sensors, operating at 313 K and 5 bar. A magnetic hotplate stirrer
with external temperature control was used to supply heat and
to stir the reaction mixtures composed by 25 mmol of acetalde-
hyde and glycerol diluted in 50% of solvent (molar basis). Amber-
lyst-15 (1.0 wt%) was used as catalyst. The experiments were
carried out for 5 h. Samples were collected at predened time
intervals and analysed in a gas chromatograph (Shimadzu, GC
2010 Plus) using a thermal conductivity detector. Helium N50
(38.5 mL min
1
) was used as the carrier gas for the separation of
the compounds in a CPWax57CB column (25 m 0.53 mm i.d.,
lmthickness of 2.0 lm). The proper column temperature program
was dened for each solvent.
2.3. Detailed kinetic study: set-up and experimental procedure
The determination of the kinetics of the acetalization of glycerol
with acetaldehyde in the presence of the selected solvent was car-
ried out in a closed glass-jacketed 1 L vessel (Buchi Laboratory
Equipment, Flawil, Switzerland), mechanically stirred and
equipped with a blow-off valve, pressure and temperature sensors.
The reactor was operated in batch mode and its temperature was
controlled through a water bath (LAUDA, LaudaKonigshofen,
Germany).
The kinetic study was performed evaluating the effect of param-
eters such as stirring speed (500 rpm and 700 rpm), temperature
(303 K, 323 K and 358 K), initial acetaldehyde to glycerol molar ra-
tio (0.5, 1.0 and 2.0), solvent molar fraction (0.25, 0.5 and 0.75) and
catalyst loading (0.15 and 0.3 wt% of the total reaction medium
mass), according to the experiments described in Table 1.
The pressure was set to 8.0 bar with helium for all the experi-
ments. Vapourliquid equilibrium calculations indicated that at
this pressure, even at the higher operating temperature, the molar
amount of the reaction medium species in the vapour phase would
not exceed 0.5% of the total molar amount introduced in the reac-
tor and, therefore, the vapour phase would be mainly constituted
by helium.
Each reaction kinetic experiment was performed only once in
order to keep the tests manageable and due to high reactants
consumption.
All the samples collected were analysed in the previously de-
scribed gas chromatograph by injecting 0.8 lL of sample (split ratio
of 15.0). Helium N50 was used as the carrier gas at a ow rate of
38.5 mL min
1
. The temperatures of the injector and thermal con-
ductivity detector were set to 573 K. The column was initially kept
at 373 K for 1 min and its temperature was then increased at a rate
of 120 K min
1
until it reached 493 K and held constant at that va-
lue for the following 6 min. After this period the column tempera-
ture was decreased to 463 K at a rate of 30 K min
1
. The total
analysis time was 60 min.
3. Results and discussion
3.1. Solvent selection for reactiveadsorptive processes
As previously stated, none of the reactants of the GEA synthesis
by means of glycerol acetalization can be used as solvent in the
SMBR technology. Therefore, a four stage methodology was devel-
oped and applied for the selection of the most suitable solvent for
the production of this acetal by reactiveadsorptive processes. For
that purpose, a typical set of operating conditions was dened and
an extensive bibliographic review was conducted in order to iden-
tify common solvents used in similar processes. The process tem-
perature was assumed to be 313 K, at a pressure of 5 bar and the
initial reaction medium composition was established as
0.25:0.25:0.50, in terms of molar fractions of acetaldehyde, glyc-
erol and solvent, respectively.
3.1.1. Theoretical based solvent screening
The list of approximately 100 solvents elaborated was reduced
considering three major process variables: solvent reactivity, sol-
vent physical state at the reaction conditions and hypothetical
phase split promoted by the use of the solvent. Solvents that have
Table 1
Kinetic experiments operating conditions.
Exp.
No.
Stirring
speed
(rpm)
Temperature
(K)
Initial
reactants
molar ratio
Solvent
molar
fraction (n/
n)
Catalyst
loading
(wt%)
1 700 323 1.00 0.50 0.15%
2 500 303 1.00 0.50 0.15%
3 500 323 1.00 0.50 0.15%
4 500 323 1.00 0.50 0.30%
5 500 343 1.00 0.50 0.30%
6 500 323 0.50 0.50 0.15%
7 500 323 2.01 0.50 0.30%
8 500 323 1.00 0.25 0.15%
9 500 323 1.00 0.75 0.30%
R.P.V. Faria et al. / Chemical Engineering Journal 233 (2013) 159167 161
been reported as reactive with any of the reactants or products of
the selected synthesis route, such as alcohols, aldehydes, ketones
and carboxylic acids, were eliminated from the primary list.
Due to operational requirements of the SMBR unit, the solvent
to be used should, preferably, be at the liquid state. Making use
of published data in the open literature the melting point of each
solvent was compiled and its boiling point determined through
the Antoine equation [44], at the previously mentioned pressure
values. Applying a safety factor of 10 K to the phase change tem-
peratures considered, the solvents for which the melting point was
above 303 K and the boiling point was below 323 K were excluded
from the screening.
In the following solvent screening step, Aspen Plus

software
was used to estimate miscibility of the solvent with the initial reac-
tion medium at the dened operating conditions, considering the
UNIFAC model for computing the activity of each species. If the
area of the miscibility region of the ternary phase diagram for a
specic solvent was lower than 75% that solvent was excluded
from the solvents list. It must be noticed that due to lack of data
for 1,3-dioxolane, chloroform, dimethyl sulfoxide and n,n-dimeth-
ylformamide, these solvents could not be excluded.
When designing an industrial process two main questions must
be considered: process performance and process impact. Regarding
the studied process, the acetalization reaction catalyzed by an ion
exchange resin in a SMBR unit, one can promptly conclude that
polarity will be a key performance parameter. On one hand, a
highly polar medium favours the acetalization reaction once its
rate determining step is the formation of a cation from the proton-
ated hemi-acetal [23], and polar media allow the stabilization of
the intermediate cations [45]. On the other hand, as an ion ex-
change resin will be used as catalyst/adsorbent, the sorption prop-
erties of the solvents can be roughly approximated by their
polarity. Water is a highly polar compound and will adsorb in large
amounts in Amberlyst-15; thus, the decrease of solvent consump-
tion (that acts as desorbent), will be favoured if a highly polar sol-
vent is chosen. In ideal terms, the polarity of the solvent should be
somewhere between the polarity of water and that of GEA. For that
reason, the solvents remaining in study from the previous pre-
selection steps were ranked (Table 2) considering a weight of
50% for process performance parameters (25% for dipole moment
and 25% for dielectric constant) and 50% for process impact param-
eters (25% for lethal dose, LD
50
, and NFPA health hazard classica-
tion; and 25% for octanolwater partition coefcient, log P
ow
, and
persistence time in evaluative environment). For each variable, a
score of 100% was attributed to the solvent that presented the
higher value for that variable and a score of 0% for the lower, if
the variable analysed is intended to be maximized. All the other
solvents were scored proportionally. The inverse scoring procedure
was applied to variables that were intended to be minimized.
In order to keep experimental tests practicable the list was
shortened, including only the solvents that had a score above
50%. However, as acrylonitrile has a severe health hazard (as can
be conrmed through its NFPA health classication maximum
hazard classication and through its Material Safety Data Sheet)
it was also excluded. Therefore, dimethyl sulfoxide, acetonitrile,
n,n-dimethylformamide were the solvents selected for further
experimental tests.
3.1.2. Experimental based solvent selection
3.1.2.1. Miscibility tests. In an attempt to assess the miscibility of
the solvents selected from the previous screening steps with the
compounds present in the reaction medium, ternary mixtures con-
taining one of the reactants, one of the products and one solvent
were prepared. Both dimethyl sulfoxide and n,n-dimethylformam-
ide have proven to be miscible with all of the reaction species. Ace-
tonitrile, however, presented some miscibility problems with
glycerol as can be observed in Fig. 1. This fact represents a handi-
cap of acetonitrile against the remaining solvents since it is ex-
pected to originate additional mass transfer resistances.
3.1.2.2. Adsorption tests. As previously mentioned, adsorption is
one of the key parameters to be analysed when dealing with SMBR
technology. It is known that the performance of these units, for in-
stance in terms of desorbent consumption, is improved if the sol-
vent presents an high adsorption capacity but preferably
between the values observed for the two products that have to
be separated. For this particular system, the adsorption capacity
of Amberlyst-15 towards the solvent should be higher than to-
wards GEA and lower than for water. The results for the swelling
tests performed are presented in Fig. 2.
As it can be concluded form Fig. 2, all the three tested solvents
present adsorption capacities between water and GEA. However,
dimethyl sulfoxide presents a slightly higher adsorption value than
the rest of the solvents. Therefore it is expected that dimethyl sulf-
oxide will lead to the minimum desorbent consumption in the
SMBR unit.
3.1.2.3. Reaction tests. The rst objective was to verify that none of
the solvents would react with any of the species of the reaction
media. Furthermore, for the assessment of the inuence of the
presence of the solvents in the kinetics of the reaction between
glycerol and acetaldehyde, experimental tests were carried out.
The results are presented in Fig. 3.
As Fig. 3 shows, dimethyl sulfoxide, n,n-dimethylformamide
and acetonitrile present similar kinetic performances, achieving
conversions of above 50% after 100 min. Moreover, no secondary
products were observed during the experiments. Although
equilibrium results are not represented in Fig. 3 (for a better
Table 2
Rank of solvents by process performance and process impact, based on their properties [44].
Solvent Dipole moment (D) Dielectric constant LD50 (mg/kg) Log P
ow
Persistence time (h) NFPA health Score (%)
Dimethyl sulfoxide 3.96 46.45 14500 1.350 573 1 82
Acetonitrile 3.92 37.50 2460 0.340 347 2 68
N,N-dimethylformamide 3.82 38.00 2800 1.010 438 2 67
Acrylonitrile 3.87 33.00 78 0.250 258 3 60
Dimethyl carbonate 0.91 3.09 13000 0.230 185 2 41
Ethyl acetate 1.78 6.02 5620 0.730 264 2 39
Methyl acetate 1.68 6.68 5000 0.180 277 2 38
1,3-Dioxolane 1.45 7.34 3000 0.370 310 2 37
Dichloromethane 1.60 9.10 1600 1.250 186 2 36
Tetrahydrofuran 1.63 7.58 1650 0.460 293 2 35
1,2-Dimethoxyethane 1.71 2.00 775 0.210 354 2 32
Chloroform 1.01 4.81 695 1.970 189 2 29
1,4-Dioxane 0.00 2.21 4200 0.270 350 2 24
Tetrachloromethane 0.00 2.20 2350 2.830 177 3 18
162 R.P.V. Faria et al. / Chemical Engineering Journal 233 (2013) 159167
understanding of the kinetic results), it was observed that n,n-
dimethylformamide has a lower equilibrium conversion (approxi-
mately 85%) than the other solvents, for which the conversion
achieved was similar (slightly above 90%), due to the effect of the
non-ideality of the mixtures.
In summary, one can say that the application of the methodol-
ogy herein developed to the production of GEA in a SMBR lead to
the identication of dimethyl sulfoxide as the most appropriate
solvent for this process, as shown by the theoretical as well as
the experimental part of the screening procedure implemented.
Besides its reduced environmental impact and health hazard, this
solvent is completely miscible with the reaction mixture in oppo-
sition to acetonitrile. On the other hand, it presents the higher
adsorption capacity, which will minimize its consumption in the
SMBR unit, without having any negative impact on the kinetics
of the reaction or the equilibrium compositions when compared
with the remaining solvents. However, a more detailed description
of the kinetics of the acetalization of glycerol in the presence of di-
methyl sulfoxide will be of signicant relevance for a correct eval-
uation of the SMBR process with this solvent. Therefore, the results
of the study of the kinetics of this reaction are presented in the
next section.
3.2. Kinetic study in the presence of the selected solvent
An experimental parametric study (summarized in Table 1) was
performed in an isothermal batch reactor in order to determine the
kinetic parameters for the acetalization of glycerol catalysed by
Amberlyst-15, in the presence of dimethyl sulfoxide.
For kinetic modelling purposes, an assessment of the mass
transfer limitations was required. Concerning mass transfer
limitations inside the catalyst, as previously reported in the open
literature [4650], a bidisperse pore size distribution, containing
micro- and macropores, was suggested for Amberlyst-15. In the
model adopted the reactants diffuse through the macropores,
adsorb in its walls and then diffuse through microspheres
(homogeneous gel particles) adsorbing in its active sites where
they react. Mass transport phenomena at the microspheres can
be considered innitely fast and, therefore, the internal mass
transfer resistance can be limited to the diffusion through the
macropores [46,48]. On the other hand, it was determined that
for the studied reaction system a stirring speed of 500 rpm could
signicantly reduce external mass transfer resistances to the point
that they became negligible.
The mathematical model implemented comprises the mass bal-
ance to the bulk and intraparticle uid given by Eqs. (1) and (2),
respectively.
dC
b;i
dt

3
R
p
1 e
b
e
b

D
eff ;i
@C
p;i
@r

rRp
1
e
p
@C
p;i
@t

1
r
2
@
@r
D
eff ;i
r
2
@C
p;i
@r

1 e
p
m
i
q
p
R 2
Species i (i = acetaldehyde, glycerol, water, GEA, dimethyl sulf-
oxide) molar concentrations in the bulk and intraparticle uid
are represented by, C
b,i
and C
p,i
, respectively; m
i
and D
eff,i
are its stoi-
chiometric and effective diffusion coefcients (computed as de-
scribed elsewhere [24]); R is the reaction rate based on the
composition along the catalyst radius, r; e
b
is the bulk porosity;
R
p
, e
p
and q
p
represent the catalyst particle radius (342.5 lm
[55]), porosity (0.36 [56]) and solid density (1016 kg m
3
[55]).
Eqs. (3)(6) dene the initial and boundary conditions consid-
ered for this model.
t 0 !C
b;i
C
b;i
0
3
t 0 !C
p;i
C
p;i
0
4
r 0 !
@C
p;i
@r
0 5
Fig. 1. Ternary phase diagram for glycerol:acetonitrile:water and glycerol:acetoni-
trile:GEA mixtures, in molar compositions at 313 K.
Fig. 2. Adsorption tests on Amberlyst-15.
Fig. 3. Comparison of the performance of the solvents on the kinetics of the
acetalization of glycerol (T = 313 K; P = 5 bar; r
acet/gly
= 1.0; x
solv
= 0.5; cat. loading of
2.0 wt% of the total reactants mass; 500 rpm).
R.P.V. Faria et al. / Chemical Engineering Journal 233 (2013) 159167 163
r R
p
!C
b;i
C
p;i

rRp
6
The reaction rate of acetaldehyde with glycerol [24], as well as
with several linear chain alcohols [5053], has been described by
several authors by a LangmuirHinshelwoodHougenWatson
model. This model assumes the adsorption of both reactants, fol-
lowed by a surface reaction that originates a hemi-acetal. Two sub-
sequent surface reactions will lead to the formation of adsorbed
water and acetal. The nal step is the desorption of both products.
The surface reaction leading to the formation of the adsorbed
hemi-acetal is considered as the rate determining step. Previous
authors have suggested that water adsorption is signicantly high-
er than that of the remaining reaction species and, consequently,
the adsorption of those species could be neglected. However, for
this system, as dimethyl sulfoxide is the major compound and it
is a very polar solvent (which indicates that it will adsorb strongly
on Amberlyst-15), its adsorption must also be considered. Eq. (7)
presents the simplied rate expression used to describe the exper-
imental data, taking into account the previous considerations.
R k
c
C
Ac
C
Gly

C
GEA
C
W
Keq

1 K
S;W
C
W
K
S;DMSO
C
DMSO

2
7
In this equation, K
eq
represents the reaction thermodynamic
equilibrium constant, k
c
, the reaction kinetic constant and K
S,W
and K
S,DMSO
the adsorption equilibrium constants for water and di-
methyl sulfoxide, respectively.
The kinetic constant depends on temperature according to
Arrhenius law (Eq. (8)),
k
c
k
c
0
exp
E
a
RT

8
where E
a
, represents the reaction activation energy and k
c
0
, is the
pre-exponential term.
The equilibrium constants depend on temperature according to
Vant Hoff law (Eqs. (9)(11)),
K
eq
exp
DG
0
RT
!
9
K
S;W
K
S;W
0
exp
DH
S;W
RT

10
K
S;DMSO
K
S;DMSO
0
exp
DH
S;DMSO
RT

11
where DG
0
, is the reaction free Gibbs energy, DH
S,W
and DH
S,DMSO
are the adsorption enthalpies for water and dimethyl sulfoxide
and K
S;W
0
and K
S;DMSO
0
, are the pre-exponential terms. From previous
work [24], computing the thermodynamic equilibrium constant in
terms of composition, a value of 12.95 kJ mol
1
was determined
for DG
0
.
The studied reaction system comprises a non-ideal mixture.
However, for modelling purposes, it was impossible to account to
this factor due to the absence in the open literature, as far as our
knowledge goes, of the necessary binary interaction parameters
between the species present in the reaction mixture. Although
the rst bibliographic references to the synthesis of GEA are dated
to 1865 [17], the study of the non-ideality of the species involved
in the different synthesis routes has been neglected by the scien-
tic community. As a consequence, no binary interaction parame-
ters involving GEA have been published so far. Without this data it
is impossible to determine the activity coefcients of each species
with the most common methods such as Wilson, NRTL or UNI-
QUAC. For the solvent free condensation of glycerol with acetalde-
hyde [24], the determination of the activity coefcients was
reported using the UNIFAC group contribution method. However,
when dimethylsulfoxide is used as solvent in this reaction, the
UNIFAC model cannot be applied since the binary interaction
parameter between the aldehyde group and dimethyl sulfoxide
does not exist in the open literature.
The model equations presented were solved with gPROMS
(General Process Modelling System, version 3.4.0) using one of its
integrated solvers, DASOLV, discretizing the radial dimension in
28 non uniform intervals and making use of a second order cen-
tered nite-difference method. The determination of the unknown
parameters, E
a
, k
c
0
, DH
S,W
, K
S;W
0
, DH
S,DMSO
and K
S;DMSO
0
, was carried
out using the Parameter Estimation tool and tting the experi-
mental data by the maximum-likelihood method, assuming a con-
stant variance model and setting the estimation tolerance to 10
5
.
For the determination of the errors associated to the parameters
estimated a Jackknife methodology [54] was applied, considering
8 groups of experiments where one experiment was removed from
the original set of experiments (experiment 1 was not taken into
account since no external mass transfer limitations were observed
and, therefore, the results are similar to those obtained in experi-
ment 3, and it was considered that the same weight should be gi-
ven to every experiment, representing a unique set of operating
conditions, on the parameter estimation procedure).
The estimation of the six parameter for the model previously
described led to large standard errors. For instance, the standard
errors for the pre-exponential factors from the Arrhenius and the
Vant Hoff laws achieved almost 100% of its absolute values. Thus,
the number of parameters to be estimated had to be reduced. Since
the range of operating temperatures can be considered narrow
(40 K), the effect of temperature on the adsorption equilibrium
was considered negligible and Eqs. (10) and (11) were removed
from the nal model. This assumption was based on the results
published for other acetalization reaction between acetaldehyde
and several alcohols [24,5153], described by a LHHW rate law,
in which it is possible to observe a more signicant temperature
dependence of the kinetic constant when compared to the adsorp-
tion constant. For instance, for the acetalization leading to the syn-
thesis of 1,1-dibutoxyethane [53], considering a temperature range
from 303 K to 343 K, the water adsorption constant is expected to
decrease to approximately one third of its value, while the increase
of the reaction kinetic constant reaches ten times its value from the
lower to the higher temperature. The parameters determined and
the corresponding errors are presented on Table 3.
The reaction rate is favoured by the increase of temperature. As
expected, Fig. 4 shows that both the experimental results, obtained
in the range of 303343 K, and the model present a higher reaction
rate for higher temperatures.
Moreover, the model has also shown to be in good agreement
with the experimental observations when changes on the initial
composition of the reaction medium were promoted. A higher
reaction rate was achieved when an excess of either of the reac-
tants was used, as presented in Fig. 5, where the rates for mixtures
with an initial acetaldehyde to glycerol molar ratio of 0.5, 1.0 and
2.0 are compared. However, the greater difculties in predicting
the effect of this parameter on the kinetic behaviour, when com-
pared with the effect of temperature, might be related with the
major impact that changes in the reaction medium composition
Table 3
Kinetic parameters estimated for the acetalization of glycerol in presence of dimethyl
sulfoxide.
Parameter (units) Estimated Value Standard Error
E
a
(kJ mol
1
) 77.7 0.8
kc
0
dm
6
mol
1
g
1
cat
s
1

5.97 10
8
1.06 10
8
K
S,W
(dm
3
mol
1
) 3.45 10
1
5.4 10
2
K
S,DMSO
(dm
3
mol
1
) 1.06 10
1
1.9 10
2
164 R.P.V. Faria et al. / Chemical Engineering Journal 233 (2013) 159167
have on the activity coefcients of the species present in the mix-
ture that could not be accounted for due to the previously men-
tioned reasons.
On the other hand, the reaction proceeded at lower rates when
the solvent amount was enlarged due to the dilution factor applied
to the reactive species (Fig. 6). Molar fractions of 0.25, 0.50 and
0.75 of dimethyl sulfoxide were studied.
The increase of the catalyst loading from 0.15 wt% to 0.30 wt%
lead to an increasing on the experimental reaction rate of approx-
imately 50% (Figs. 46), estimated by the initial reaction rate meth-
od. A similar increase was observed for the model prediction.
The parametric study performed showed that the model is able
to predict the effect of changes in the design variables. In order to
make a more objective assessment of the quality of the tting of
the experiments, considering the parameters estimated for the
mathematical model implemented, its correlation coefcient was
computed (Eq. (12)). The value determined was 0.980, indicating
that the model describes with high accuracy the kinetic behaviour
of this reaction over the wide range of experimental conditions
tested.
R
2
1
P
NE
i1
P
NP
j1
X
exp
i;j
X
mod
i;j

2
P
NE
i1
P
NP
j1
X
exp
i;j
X

exp

2
12
To conclude, the values obtained for the kinetic parameters of the
reaction for the synthesis of GEA in the presence of dimethyl
sulfoxide were compared with the values obtained in a previous
work for the solvent free reaction, for which the authors have deter-
mined an activation energy of 51.2 kJ mol
1
[24]. The difference to
the value reported hereby (approximately 35%) may rely on several
factors. First of all, the prior model (in the absence of solvent) has
been described in terms of activities [24], while the model for this
new system (in the presence of dimethyl sulfoxide) does not ac-
count for the non ideality of the liquid phase, due to lack of param-
eters for the computation of the activity coefcients (namely, the
binary interaction coefcient between dimethyl sulfoxide and the
aldehyde group of the UNIFAC method). Additionally, the inuence
of the temperature on the adsorption equilibrium was not consid-
ered for the reaction in the presence of the solvent as previously de-
scribed, since the number of parameters to be estimated would not
allow a reliable determination of their values. As a consequence, the
enthalpies of adsorption, and their impact on the adsorption con-
stants and in the reaction rate according to the LangmuirHinshel-
woodHougenWatson model, might have a slight inuence on the
activation energy value. Finally, the high polarity of dimethyl sulf-
oxide and its corresponding afnity towards the resin, where it is
expected to adsorb strongly, might be leading to the hindering of
some of the available active sites of the catalyst, decreasing the
overall reaction rate. The direct comparison between the adsorption
constant values here reported and those for the solvent free system
is not allowed. The two models have different considerations over
the non-ideality of the mixture and the values estimated for these
constants are affected by them. However, it is important to mention
that, as expected due to the well-known high afnity of water to-
wards ion exchange resins, as Amberlyst-15, a higher adsorption
capacity was estimated for water than for dimethyl sulfoxide. Nev-
ertheless, it is important to underline that no direct measurement
of the adsorption constants was performed. Instead, the values pre-
sented were obtained through the tting of kinetic data. In any case,
it is intended that those values are coherent between them and, if
not accurate, at least they should have some physical meaning.
The simulation of the concentration proles along the catalyst
particles radius allowed the determination of the reaction effec-
tiveness factor, for which a value of 57% was obtained when eval-
uated at 343 K (the highest temperature assessed, at which higher
inuence of the mass transfer resistances is expected) and consid-
ering the average particle diameter of commercial Amberlyst-15,
685 lm.
4. Conclusions
Although solvent selection methodologies for different applica-
tions have been the aim of several research works, the authors of
Fig. 4. Effect of the temperature on the reaction rate (P = 8.0 bar; r
acet/gly
= 1.0;
x
solv
= 0.5 (n/n); stirring speed of 500 rpm). The full lines represent the conversion
estimated by the model.
Fig. 5. Effect of the reactants molar ratio on the reaction rate (P = 8.0 bar; T = 323 K;
x
solv
= 0.5 (n/n); stirring speed of 500 rpm). The full lines represent the conversion
estimated by the model. The full lines represent the conversion estimated by the
model.
Fig. 6. Effect of the solvents molar fraction on the reaction rate (P = 8.0 bar;
T = 323 K; r
acet/gly
= 1.0; stirring speed of 500 rpm). The full lines represent the
conversion estimated by the model.
R.P.V. Faria et al. / Chemical Engineering Journal 233 (2013) 159167 165
this study have perceived that for integrated reactiveadsorptive
processes as SMBR, the information on this topic was scarce and
therefore there was a need to develop a concise and prompt meth-
odology for this purpose. The methodology suggested in this paper
comprises a two stage screening. First, an extensive list of the most
commonly used solvents in chemical processes is reduced based on
their properties, reported in the literature. Then, experimental
tests are performed with the solvents selected from the previous
screening step. When the methodology developed was applied to
the synthesis of GEA, through the acetalization of glycerol, cata-
lyzed by Amberlyst-15, dimethyl sulfoxide has demonstrated to
be the most suitable solvent for this reaction, presenting top per-
formances on almost every topic. It proved to be unreactive and
miscible with the reaction medium and most importantly, it pre-
sented an adsorption capacity for the catalyst/adsorbent of the pro-
cess that can predictably reduce the solvent consumption on a
future SMBR unit. Moreover, its use has low health and environ-
mental impact.
Finally, a detailed study of the kinetic of the acetalization of
glycerol was conducted. The mathematical model developed was
able to predict with high accuracy the effect on the reaction rate
promoted by changes in the design variables, with a correlation
coefcient of 0.980. The kinetic law was described by the Lang-
muirHinshelwoodHougenWatson method. A value of
77.7 kJ mol
1
was estimated for its activation energy. When com-
pared with the solvent free reaction, the presence of dimethyl sulf-
oxide appears to have some impact on the reaction performance,
consequence of the dilution factor and the reduction of available
catalyst active sites due to its adsorption.
Acknowledgments
The research leading to these results received funding from the
European Union Seventh Framework Programme (FP7/2007-2013)
under Grant Agreement 241718 EuroBioRef. This work was also
nancially supported by Fundao para a Cincia e a Tecnologia,
Project Grant PTDC/EQU-EQU/100564/2008. C.P. gratefully
acknowledges nancial support from Fundao para a Cincia e a
Tecnologia, Postdoctoral Research Fellowship SFRH/BPD/71358/
2010.
References
[1] G. Sorda, M. Banse, C. Kemfert, An overview of biofuel policies across the
world, Energy Policy 38 (2010) 69776988.
[2] A. Behr, J. Eilting, K. Irawadi, J. Leschinski, F. Lindner, Improved utilisation of
renewable resources: new important derivatives of glycerol, Green Chem. 10
(2008) 1330.
[3] C.J.A. Mota, C.X.A.d. Silva, V.L.C. Gonalves, Gliceroqumica: novos produtos e
processos a partir da glicerina de produo de biodiesel, Quim. Nova 32 (2009)
639648.
[4] C.H. Zhou, J.N. Beltramini, Y.X. Fan, G.Q. Lu, Chemoselective catalytic
conversion of glycerol as a biorenewable source to valuable commodity
chemicals, Chem. Soc. Rev. 37 (2008) 527549.
[5] H. Seram, I.M. Fonseca, A.M. Ramos, J. Vital, J.E. Castanheiro, Valorization of
glycerol into fuel additives over zeolites as catalysts, Chem. Eng. J. 178 (2011)
291296.
[6] I. Agirre, I. Garca, J. Requies, V.L. Barrio, M.B. Gemez, J.F. Cambra, P.L. Arias,
Glycerol acetals, kinetic study of the reaction between glycerol and
formaldehyde, Biomass Bioenergy 35 (2011) 36363642.
[7] A. Jaecker-Voirol, I. Durand, G. Hillion, B. Delfort, X. Montagne, Glycerin for
new biodiesel formulation, Oil Gas Sci. Technol. 63 (2008) 395404.
[8] B. Delfort, I. Durand, A. Jaecker-Voirol, T. Lacome, X. Montagne, F. Paille, Diesel
Fuel Compounds Containing Glycerol Acetals; US 2003/0163949, in, 2003.
[9] G.W. Mushrush, S.C. Basak, J.E. Slone, E.J. Beal, S. Basu, W.M. Stalick, D.R. Hardy,
Computational study of the environmental fate of selected aircraft fuel system
deicing compounds, J. Environ. Sci. Health Part A Toxic/Hazard. Substances
Environ. Eng. 32 (1997) 22012211.
[10] G.W. Mushrush, E.J. Beal, D.R. Hardy, J.M. Hughes, J.C. Cummings, Jet fuel
system icing inhibitors: synthesis and characterization, Ind. Eng. Chem. Res. 38
(1999) 24972502.
[11] G.W. Mushrush, E.J. Beal, D.R. Hardy, W.M. Stalick, S. Basu, D. Grosjean, J.
Cummings, Jet fuel system icing inhibitors: synthesis and stability, ACS Div.
Fuel Chem., Preprints 43 (1998) 6062.
[12] G.W. Mushrush, W.M. Stalick, E.J. Beal, S.C. Basu, J. Eric Slone, J. Cummings, The
synthesis of acetals and ketals of the reduced sugar mannose as fuel system
icing inhibitors, Pet. Sci. Technol. 15 (1997) 237244.
[13] E. Garca, M. Laca, E. Prez, A. Garrido, J. Peinado, New class of acetal derived
from glycerin as a biodiesel fuel component, Energy Fuels 22 (2008) 4274
4280.
[14] S.D. Varfolomeev, G.A. Nikiforov, V.B. Volieva, G.G. Makarov, L.I. Trusov, Agent
for Increasing the Octane Number of a Gasoline Automotible Fuel; WO/2009/
145674, 2009.
[15] J.M. Delgado Puche, Procedure to obtain biodiesel fuel with improved
properties at low temperature; US 7 637 969, Industrial Management S.A.,
Madrid, 2009.
[16] G.H. Hillion, B.P. Delfort, I.R.M. Durand, Method For Producing Biofuels,
Transforming Triglycerides Into At Least Two Biofuel Families: Fatty Acid
Monoesters And Ethers and/or Soluble Glycerol Acetals; US 2007/0283619 A1,
2007.
[17] T. Harnitzky, N. Menschutkine, Ueber die Verbindungen des Glycerins mit den
Aldehyden, Justus Liebigs Annalen der Chemie 136 (1865) 4.
[18] J.U. Nef, Dissociationsvorgnge in der Glycol-Glycerinreihe, Justus Liebigs
Annalen der Chemie 335 (1904) 191245.
[19] H.S. Hill, H. Hibbert, Studies on reactions relating to carbohydrates and
polysaccharides V. The use of acetylene for the synthesis of cyclic acetals, J.
Am. Chem. Soc. 45 (1923) 31083116.
[20] H.S. Hill, A.C. Hill, H. Hibbert, Studies on the reactinos relating to
carbohydrates and polysaccharides. XVI. Separation and Identication of the
isomeric ethylidene glycerols, J. Am. Chem. Soc. 50 (1928) 22422249.
[21] M. Martin Maglio, C.A. Burger, Condensation reactions: a laboratory
preparation of ethylidene glycerol, J. Chem. Educ. 23 (1946) 174175.
[22] A.S. Shevchuk, V.A. Podgornova, E.G. Piskaikina, Features of synthesis of 4-
hydroxymethyl-1,3-dioxolanes from glycerol and aldehydes, Russ. J. Appl.
Chem. 72 (1999) 853856.
[23] A.C. Da Silva Ferreira, J.C. Barbe, A. Bertrand, Heterocyclic acetals from glycerol
and acetaldehyde in port wines: evolution with aging, J. Agric. Food. Chem. 50
(2002) 25602564.
[24] R.P.V. Faria, C.S.M. Pereira, V.M.T.M. Silva, J.M. Loureiro, A.E. Rodrigues,
Glycerol valorization as biofuel: thermodynamic and kinetic study of the
acetalization of glycerol with acetaldehyde, Ind. Eng. Chem. Res. 52 (2013)
15381547.
[25] D.J. Miller, L. Peereboom, A.K. Kolah, N.S. Asthana, C.T. Lira, Process for
production of a composition useful as a fuel; US2006/0199970 A1, United
States, 2008.
[26] X. Hong, O. McGiveron, C.T. Lira, A. Orjuela, L. Peereboom, D.J. Miller, A reactive
distillation process to produce 5-hydroxy-2-methyl-1,3-dioxane from mixed
glycerol acetal isomers, Org. Process Res. Dev (2011).
[27] X. Hong, O. McGiveron, A.K. Kolah, A. Orjuela, L. Peereboom, C.T. Lira, D.J.
Miller, Reaction kinetics of glycerol acetal formation via transacetalization
with 1,1-diethoxyethane, Chem. Eng. J. 222 (2013) 374381.
[28] X. Hong, A.K. Kolah, C.T. Lira, D.J. Miller, An improved approach for cyclic
acetals from glycerol, Nashville, TN, 2009.
[29] A.E. Rodrigues, C.S.M. Pereira, J.C. Santos, Chromatographic reactors, Chem.
Eng. Technol. 35 (2012) 11711183.
[30] C.S.M. Pereira, P.S. Gomes, G.K. Gandi, V.M.T.M. Silva, A.E. Rodrigues,
Multifunctional reactor for the synthesis of dimethylacetal, Ind. Eng. Chem.
Res. 47 (2008) 35153524.
[31] A.E. Rodrigues, V.M.M.D. Silva, Industrial Process for Acetals Production in a
Simulated Moving Bed Reactor; WO/2005/11347, 2005.
[32] V.M.T.M. Silva, A.E. Rodrigues, Novel process for diethylacetal synthesis, AlChE
J. 51 (2005) 27522768.
[33] N.S. Graa, L.S. Pais, V.M.T.M. Silva, A.E. Rodrigues, Analysis of the synthesis of
1,1-dibutoxyethane in a simulated moving-bed adsorptive reactor, Chem. Eng.
Process.: Process Intensication 50 (2011) 12141225.
[34] S. Luo, J.L. Falconer, Acetone and acetaldehyde oligomerization on TiO
2
surfaces, J. Catal. 185 (1999) 393407.
[35] R. Gani, C. Jimnez-Gonzlez, D.J.C. Constable, Method for selection of solvents
for promotion of organic reactions, Comput. Chem. Eng. 29 (2005) 16611676.
[36] M. Folic , R. Gani, C. Jimnez-Gonzlez, D.J.C. Constable, Systematic selection of
green solvents for organic reacting systems, Chin. J. Chem. Eng. 16 (2008) 376
383.
[37] S. Perez-Vega, E. Ortega-Rivas, I. Salmeron-Ochoa, P.N. Sharratt, A system view
of solvent selection in the pharmaceutical industry: towards a sustainable
choice, Environ. Dev. Sustain. 15 (2013) 121.
[38] A.I. Papadopoulos, P. Linke, Integrated solvent and process selection for
separation and reactive separation systems, Chem. Eng. Process.: Process
Intensication 48 (2009) 10471060.
[39] J.P. Taygerly, L.M. Miller, A. Yee, E.A. Peterson, A convenient guide to help
select replacement solvents for dichloromethane in chromatography, Green
Chem. 14 (2012) 30203025.
[40] R. Hu, Y. Pan, Recent trends in counter-current chromatography, TrAC Trends
Anal. Chem. 40 (2012) 1527.
[41] C. Jimnez-Gonzalez, A.D. Curzons, D.J.C. Constable, V.L. Cunningham,
Expanding GSKs Solvent selection guide application of life cycle
assessment to enhance solvent selections, Clean Technol. Environ. Policy 7
(2004) 4250.
166 R.P.V. Faria et al. / Chemical Engineering Journal 233 (2013) 159167
[42] K. Alfonsi, J. Colberg, P.J. Dunn, T. Fevig, S. Jennings, T.A. Johnson, H.P. Kleine, C.
Knight, M.A. Nagy, D.A. Perry, M. Stefaniak, Green chemistry tools to inuence
a medicinal chemistry and research chemistry based organisation, Green
Chem. 10 (2008) 3136.
[43] P.G. Jessop, Searching for green solvents, Green Chem. 13 (2011) 13911398.
[44] C.L. Yaws, Yaws Handbook of Physical Properties for Hydrocarbons and
Chemicals, 2008.
[45] G. Wypych, Handbook of Solvents, ChemTec Publishing, 2001.
[46] R.M. Quinta Ferreira, C.A. Almeida-Costa, A.E. Rodrigues, Heterogeneous
models of tubular reactors packed with ion-exchange resins: simulation of
the MTBE synthesis, Ind. Eng. Chem. Res. 35 (1996) 38273841.
[47] S.-K. Ihm, J.-H. Ahn, Y.-D. Jo, Interaction of Reaction and Mass Transfer in Ion-
Exchange Resin Catalysts, Ind. Eng. Chem. Res. 35 (1996) 29462954.
[48] T. Dogu, E. Aydin, N. Boz, K. Murtezaoglu, G. Dogu, Diffusion resistances and
contribution of surface diffusion in TAME and TAEE production using
Amberlyst-15, Int. J. Chem. React. Eng. 1 (2003).
[49] K. Sundmacher, R.-S. Zhang, U. Hoffmann, Mass transfer effects on kinetics of
nonideal liquid phase ethyl tert-butyl ether formation, Chem. Eng. Technol. 18
(1995) 269277.
[50] V.M.T.M. Silva, A.E. Rodrigues, Kinetic studies in a batch reactor using ion
exchange resin catalysts for oxygenates production: role of mass transfer
mechanisms, Chem. Eng. Sci. 61 (2006) 316331.
[51] G.K. Gandi, V.M.T.M. Silva, A.E. Rodrigues, Process development for
dimethylacetal synthesis: thermodynamics and reaction kinetics, Ind. Eng.
Chem. Res. 44 (2005) 72877297.
[52] V.M.T.M. Silva, A.r.E. Rodrigues, Synthesis of diethylacetal: thermodynamic
and kinetic studies, Chem. Eng. Sci. 56 (2001) 12551263.
[53] N.S. Graa, L.S. Pais, V.M.T.M. Silva, A.E. Rodrigues, Oxygenated biofuels from
butanol for diesel blends: synthesis of the acetal 1,1-dibutoxyethane catalyzed
by amberlyst-15 ion-exchange resin, Ind. Eng. Chem. Res. 49 (2010) 6763
6771.
[54] R.G. Miller, The jackknife: a review, Biometrika 61 (1974) 115.
[55] C.S.M. Pereira, M. Zabka, V.M.T.M. Silva, A.E. Rodrigues, A novel process for the
ethyl lactate synthesis in a simulated moving bed reactor, Chem. Eng. Sci. 64
(2009).
[56] F. Lode, M. Houmard, C. Migliorini, M. Mazzotti, M. Morbidelli, Continuous
reactive chromatography, Chem. Eng. Sci. 56 (2001) 269291.
R.P.V. Faria et al. / Chemical Engineering Journal 233 (2013) 159167 167