Chemical Engineering and Processing 96 (2015) 1–13

Contents lists available at ScienceDirect

Chemical Engineering and Processing:

Process Intensification
journal homepage: www.elsevier.com/locate/cep

Comparison of different reactive distillation schemes for ethyl acetate

production using sustainability indicators
Miguel A. Santaella, Alvaro Orjuela* , Paulo C. Narváez
Grupo de Procesos Químicos y Bioquímicos, Departamento de Ingeniería Química y Ambiental, Universidad Nacional de Colombia Sede Bogotá, Cra 30 # 45-
03, Edificios 412, Bogotá, Colombia

A R T I C L E I N F O A B S T R A C T

Article history: An assessment of different intensified processes for ethyl acetate production by direct esterification was
Received 30 March 2015 performed. After the selection and validation of the adequate thermodynamic and kinetic models, a
Received in revised form 3 July 2015 comparison among different reported processing technologies (i.e., conventional, reactive distillation,
Accepted 30 July 2015
reactive distillation with pressure swing, dividing wall column with reactive reboiler), and a novel
Available online 3 August 2015
proposed configuration using a reactive dividing wall column was carried out. Specifications of raw
materials and products used as constrains during simulations were defined according with market
Keywords:
requirements. To perform a fair assessment among different alternatives, each process was optimized
Reactive distillation
Dividing wall column
using a mixed strategy of sensitivity analysis coupled with sequential quadratic programming algorithm.
Ethyl acetate Total annual cost was selected as the optimization variable. Conversion, productivity, mass intensity,
Esterification mass productivity, Sheldon’s Factor (E-Factor), water-free Sheldon’s factor (EW-Factor), and energy
Sustainability intensity were used as sustainability indicators. According with results, the novel reactive dividing wall
column configuration proposed in this work ends up being the more energy efficient and cost effective
(46% energy and 26% cost savings compared with the traditional process) and also it is characterized by
the best sustainability indicators.
ã 2015 Elsevier B.V. All rights reserved.

1. Introduction Among the different types of chemicals currently produced

Over the last decades, there has been an increasing interest in
products and processes that meet sustainable development
criteria. After the broad definition of sustainability accounting
for the triple bottom line dimensions (environmental, social and
economic), different attempts have been done to transform such a
concept into basic green engineering principles, and furthermore
into sustainability indicators to measure how green processes or
products can be. Despite the sustainability indicators can be used
to evaluate existing processes, its major value can be encountered
when applied in the analysis of new processes that need to be more
cost-effective and environmentally friendly than existing ones; for
example, when developing processes that switch from fossil to bio-
based feedstock. Due to the increase use of biomass as a feedstock,
and taking into account its higher costs compared with fossil
feedstock, it is necessary to develop more efficient technologies to
reach cleaner, safer, and more cost-effective production of bio-
based derivatives [1].
* Corresponding author.
E-mail address: aorjuelal@unal.edu.co (A. Orjuela).
from non-renewable feedstock that can be replaced by bio-based
alternatives, solvents stand out because their intensive consump-
tion in a large variety of processes and products. In addition to the
fossil resources depletion, the environmental impacts of using
traditional solvents are evident, for instance in the contamination
of surface and underground waters and also in the air pollution
near urban areas. From the variety of traditional solvents, ethyl
acetate (EtAc) distinguishes as a major industrial commodity,
which is mainly used as solvent for paints, coatings, resins, inks,
and in decaffeination. It is also a main ingredient in different
fragrances and flavors for consumer products.
There are several chemical routes for EtAc production
implemented at the industrial scale, using both fossil-based and
bio-based raw materials (Table 1). EtAc is mainly produced by
esterification of acetic acid and ethanol in liquid or vapor phase (1),
by acetylation of ethylene (2), and by ethanol dehydrogenation (3).
The safety indicators [2] and atom economies of these routes are
also presented in Table 1.
The presence of ethylene in the olefin acetylation (route 2), the
production of hydrogen in the Tishchenko pathway (route 3)
together with the extreme conditions of both processes affect
negatively their safety indicators. This reduces the benign-ness of

http://dx.doi.org/10.1016/j.cep.2015.07.027
0255-2701/ ã 2015 Elsevier B.V. All rights reserved.
2 M.A. Santaella et al. / Chemical Engineering and Processing 96 (2015) 1–13
Nomenclature
BRD Batch reactive distillation
C Concentration (mol/L)
Ea Activation energy (J/mol)
EtAc Ethyl acetate (–)
EtOH Ethanol (–)
H2O Water (–)
H2SO4 Sulfuric acid (–)
HAc Acetic acid (–)
k Kinetic constant (mol/s)
Kc Equilibrium constant
PS Pressure swing
RD Reactive distillation (–)
RDWC Reactive dividing wall column (–)
Wcat Catalyst mass fraction (kgcatalyst/kgsolution)
x Mole fraction liquid phase (–)
y Mole fraction vapor phase (–)
TAC Total annual cost (million Usd/year)
these routes despite the fact both have high atom economy and
that hydrogen co-produced in the second route could be of interest.
Comparatively, direct esterification (route 1) has the lowest atom
economy from all three reactions; however it has the lower value
of the safety indicator being the most benign chemical route
according with metrics described by Srinivasan et al. [2].
Additionally, in a recent study it was found that among the
chemical routes of Table 1, direct esterification of bio-based acetic
acid and ethanol appears to be the more cost-effective alternative
as well as the one with the lower global warming potential [3].
Despite Fisher esterification to EtAc is a classic example used in
the academia to introduce basic concepts on chemical reaction
engineering and thermochemistry, this system is still under active
investigation. In recent years, different processing technologies
applied to EtAc production have been reported, most of which
focus on a process intensification (PI) approach. Notwithstanding
the proposed technologies have claimed to be cost-effective and
environmentally friendly, no sustainability assessment of such
technologies has been carried out. Furthermore, as there is room
for future processes development by introducing more complexity
during PI (for instance by implementing heat-integrated reactive
separations), there is need for including also the sustainability
evaluation in the early stages of the process synthesis.
Benefits of PI are associated with the enhanced safety, space and
waste reductions, energy savings, higher efficiency and productiv-
ity, and consequently better economic and environmental perfor-
mance. As an effort to classify most PI attempts reported in the
literature, Van Gerven and Stankiewick [4] described four
approaches commonly used according with the studied domain:
thermodynamic, functional, spatial, and temporal.
From a general point of view, the thermodynamic approach for PI
consists of energy integration of streams, stages or operations, to
replace the required amount of heating and/or cooling provided with
Table 1
Industrial chemical routes for ethyl acetate production and some sustainability indicators.
Chemical route Reaction
1 CH3 CH2 OH þ CH3 COOHþ$CH3 COOCH2 CH3 þH2 O
2 CH2 ¼ CH2 þCH3 COOH ! CH3 COOCH2 CH3
3 2CH3 CH2 OH ! CH3 COOCH2 CH3 þ2H2
a
Calculated as a cumulative index of the four chemical safety parameters (Toxicity, Reactivity, Explosiveness and flammability) as mentioned in [2].
utilities. The use of preheating exchangers, pump-arounds, vapor
recompression, and dividing wall columns are examples of such
energy integrations. The functional domain deals with the integra-
tion of two or more operations or technologies within a single unit.
This is the case of hybrid and reactive separations such as reactive
distillation (RD), catalytic distillation, reactive adsorption, mem-
brane distillation, and membrane reactors among others.
Considering all the above, this work studies and develops a novel
processing technology for EtAc production by using a reactive
dividing wall column. Although RD and dividing wall column with
reactive reboiler has already been proposed separately [5,6], the
proposed configuration of using a reactive dividing wall column
(RDWC) is innovative for EtAc production, and has not been reported
in the open literature. The performance of the proposed technology
was compared with other intensified processes by considering
quantitative economic and sustainability indicators.
Because most reports on EtAc production by intensified
processes have been developed using different modeling basis
(thermodynamics, kinetics, models, etc.), specifications (raw
materials and product purities, catalysts, etc.), and assumptions;
a direct comparison among them is difficult and anyhow
subjective. In this sense, it was necessary to carry out such
comparison under equivalent conditions with the optimized
processes. To the knowledge of the authors, this comparison of
current intensified technologies (including economic and sustain-
ability indicators) has not been previously accomplished.
2. Methodology
In order to objectively compare the different intensified
processing alternatives reported in the literature, a complete
revision of the fundamentals, constrains, and specifications of the
process was necessary. To begin with, a proper thermodynamic
model that adequately describes phase equilibria of the reactive
systems was selected. Regarding reaction kinetics, there was the
need to select a specific catalyst and to explore and validate the
corresponding kinetic model. Afterwards, a review of the reported
intensified processes for EtAc production was performed, paying
special attention to reactive distillation sequences. Detailed
described processing alternatives were selected for simulations,
and further sustainability analysis was performed. Simulation of
the different schemes was performed by mean of steady-state
equilibrium-stage simulations using Aspen plus 7.31. Raw
materials and product specifications used during simulations
were the same for all intensified processes under study. In this
sense, specifications were established according with commercial
technical datasheets. In some cases, some adjustments over the
original reported processes were necessary to achieve the
specifications of the commercial product. The design pressure
for all columns was selected to be 1 bar, except for the pressured
column in the pressure swing process. Temperature in the
condensers and decanters was verified to guarantee the use of
water from cooling towers, which was assumed to be at 25
C.
After all simulations met the design criteria, the total annual
costs (TAC) were computed for each configuration. To do so, a 3-
year period for return on investment was used. Since world’s
Atom economy (%) Safety indicatora
83,0 1.7
100 2.6
95,6 2.2
 M.A. Santaella et al. / Chemical Engineering and Processing 96 (2015) 1–13 3

market for EtAc is around 3 million tons per year, and large plants
have a capacity for manufacturing around 100.000 t per year [7],
Total mass fed ðkgÞ  Water ðkgÞ
this was selected as the base case production capacity. For the fixed mass IntesityðMIÞ ¼ (11)
mass of product ðkgÞ
costs calculation, the methodology proposed by Douglas [8] was
used (Eqs. (1)–(4). To calculate the variable costs, updated
averaged raw material and utility prices were consulted. Natural
1
gas was used as fuel and 85% efficiency was assumed for heating Mass productivityðMPÞ ¼  100 (12)
MI
loop.
TAC ¼ Fixed Costs þ Variable Costs (1) The EI factor measures the amount of energy used per kilogram
of product. In this case, we added up the major sources of power
consumption (i.e., distillation) and then dividing it by the mass
Installed Cost flow of pure EtAc. Sheldon’s E-Factor is a measure of the adequate
Fixed Costs ¼ (2)
3yr or inadequate utilization of resources; it quantifies the mass of
Installed Costs ¼ ðBase CostÞðCost indexÞðIF þ Fc  1Þ (3) waste generated per kilogram of product. For some authors, all
effluents from the process different from the main product stream
here, IF corresponds to installation factor, and Fc is a correction will be considered waste while other authors consider water
factor for material, pressure, etc. [8]. Operating costs were should not be included in the total waste calculation. We have
calculated based upon utilities consumption, specifically heating included both calculations since water is used in some processes to
costs. facilitate phase splitting, but it is also generated as a byproduct of
 
USD the reaction. An analysis regarding the differences between both
EnergyCosts ¼ Sheldon's factors (water-free and not water-free) will be carried
yr

    out in the discussion section with the calculated values from each
Heat duty kWh
 Natural gas price USD
m3
 8160 yr
h
alternative. The MI factor is a ratio of the total mass required per kg
  (4) of product; a MI value of 1 means that all of the mass consumed in
0; 85  Natural gas energy kW m3
the process is being transformed into useful product. The greater
the value of the MI factor, the greater the amount of byproducts
After the preliminary economic evaluation of the base cases, and waste. MI does not take into account the amount of water
optimization of each configuration was performed by running a being used since it is considered as a carrier or solvent rather than
sensitivity analysis over the discrete variables, coupled with a as a reactant, likewise it is also not taken into account as part of a
sequential quadratic programing (SQP) algorithm for the continu- product. The MP, is the mass percentage that is actually being
ous variables to minimize TAC. Thereafter, sustainability indicators transformed into useful product, it may be calculated as the inverse
were calculated to complement the economical comparison. First, of the MI.[10]
traditional indicators such as conversion, recovery and productivi-
ty were calculated according to Eqs. (5) and (7). These indicators 3. Modelling fundamentals
contribute to evaluate the inherent safety and therefore the
sustainability of a given process since conversion and recovery, and 3.1. Thermodynamics
hence productivity and yield, are also directly related to the
inventory of reactants and recycle streams flow rates.[2] The quaternary reactive system under study (HAc–EtOH–EtAc–
H2O) exhibits non-ideal behavior with the formation of three
moles of HAC Converted homogeneous azeotropes and one heterogeneous azeotrope [11].
ConversionðXÞ ¼ (5)
moles of HAC Fed Additionally, H2O and EtAc are partially immiscible; therefore,
various activity coefficient models have been proposed to address
the non-ideality of the liquid phase. Regarding the vapor phase,
moles of EtAc in product Stream
RecoveryðRcÞ ¼ (6) acetic acid has been found to dimerize, and this behavior has been
moles of HAC Converted
successfully modeled with equations of state corrected by the
chemical theory (e.g., Hayden O’Connell—HOC). Nevertheless,
ProductivityðPÞ ¼ X  Rc some previous researchers have assumed ideality in the vapor
moles of EtAc in product Stream phase to simplify the VLLE model. In a previous work, Tang et al. [5]
¼ (7) validated a set of parameter for an NRTL-HOC model with
moles of HAC Fed
experimental data of pure, binary and ternary mixtures. Because
the goodness of the model, this has been applied in subsequent
Then, energy intensity (EI), Sheldon’s E-factor, water-free investigations [12–14], and will be used in this work to describe
Sheldon’s EW-Factor mass intensity (MI), and mass productivity phase equilibrium of the reactive system. Parameters of the
(MP) [9,10], were calculated according to Eqs. (8)–(12). thermodynamic model are summarized in the supplementary
Energy usedðWÞ material.
Energy intensityðEIÞ ¼ (8)
mass of product ðkgÞ
3.2. Kinetic models

Total waste streams ðkgÞ A large amount of kinetic studies on Fisher esterification to EtAc
E¼ (9) have been reported using acid catalysts. Traditionally, H2SO4 has
mass of product ðkgÞ
been widely used as catalyst because is more active than most
Total waste streams ðkgÞ  Water in waste streamsðkgÞ
EW ¼ commercial homogeneous (e.g., p-toluene sulfonic acid, Meth-
mass of product ðkgÞ
anesulfonic acid) and heterogeneous (e.g., ion exchange resins)
(10) catalysts. H2SO4 loadings in commercial applications range from
0.2 to 1% wt of the reactive mixture [15]. However, there are some
 4
Table 2
Kinetic expressions reported for acetic acid–ethanol esterification with different catalyst.

Catalyst Rate Parameters Ref Units Comments Modifications

 
H2SO4—1 k1 6500:1 [23] r:mol/sm3, C:mol/
r 1 ¼ k1 C 1 C 2  C 3 C 4 k1 ¼ ð4:195C k þ 0:08815Þexp 
Kc T m3
Ck:%vol
kc ¼ 7:558  0:012T
 
H2SO4—2 dC AcA 310 [19] C:Sulphuric acid C units changed to weight
rA ¼ ¼ k1 C nAcA C m p q
EtA  k2 C EtAc C H2 O k1 ¼ 4:86  104 þ 0:496exp  þ 0:205C 0:85
dt T conc (wt%) fraction
 
220 rA:kmol/m3min
k2 ¼ 3::715  104 þ 0:01572exp  þ 0:09785C 0:85
T Ci:kmol/m3
n = m = 0.5, p = q = 1
HCl r ¼ k1 c1 c2  k2 c3 c4 k1 = 4.76  104 [21] C:mol/L “K” is non temperature dependent
k2 = 1.63  104 r:gmol/(Lmin)
T = 100
C k:L/(mingmol)

 
keq ¼ 4:0  0:2
Amberlyst 2:6  1014 e 104129 C A C B  CEKCW [22] Ci:mol/dm3 Forward frequency factor units Authors used mole
dC RT
rA ¼  A ¼

M.A. Santaella et al. / Chemical Engineering and Processing 96 (2015) 1–13

15 – 1 Time :min (m3)2 kmol2 s1 are not consistent with fraction to fit kinetic
dt ðC B þ 3:7C W Þ
Cinetic factor concentration based kinetics model. Ci replaced by xi
    (m3)2kmol2 s1

Amberlyst Ea;3 xEtAc xH2 O Ka,3 = 1320 [24] kmol 1 Validated but non peer reviewed
r3 ¼ wcat rsol ka;3 exp  xAcAc xEtOH  ka;3 :
15–2 RT K a;3 Ka,3 = 4550 kgcat s
  Ka,3 = 2.71 Ea,3:kJ/kmol
Amberlyst dC k K = 0.0026 [20] k:L/(molmin) “k” units are not mentioned, although it
 A ¼ kC A C B  CE CW
15–3 dt Ke Ke = 3.2 should be in L/(molmin). It is non
T = 343K (70
C) temperature dependent
Wcat = 100gcat/L  
Amberlyst r ¼ mcat ðk1 C HAc C EtOH  k1 C EtAc C H2 O Þ 6105:6 [14] m:gcat
k1 ¼ 1:24  109 exp 
35 wet T r:mol/min
 
5692:1
k1 ¼ 1:34  108 exp 
T
 
Amberlyst r ¼ wcat ðk þ C HAc C EtOH  k1 C EtAc C H2 O Þ 57960 [13] R: Jmol/K “R” should be in J/(molK) instead of Jmol/
kþ ¼ 131137exp 
36 RT C: mol/L K
  r:mol/Lgmin “r” includes Wcat in expression, therefore
60550 k:L/molming units should be mol/Lmin instead of mol/
k ¼ 81389exp 
RT wcat:g Lgmin
      
Amberlyst Ea;3 xEtAc xH2 O ka,3 = 12,500  200 [17] kmol 1
r3 ¼ wcat rsol ka;3 exp  xAcAc xEtOH  ka;3 :
70 RT K a;3 Ea,3 = 56,700  200 kgcat s
Ka,3 = 3.0  0.1 Ea,3:kJ/kmol
Purolite1 r ¼ k1 xEtOH xm
AcOH  k2 xEtOAc xH2 O m = 1.5 [25] k1,k2:mol/kgcats
 
CT179 48300 R: J/mol
k1 ¼ 4:24  106 exp 
RT  
66200
k2 ¼ 4:55  108 exp 
RT
 
Auto cat.— r ¼ k1 c1 c2  k2 c3 c4 59774 [23] Ci: mol/dm3
k1 ¼ 0:485exp 
1 RT r: mol/sm3
  k: m3/mols
59774
k2 ¼ 0:123exp 
RT
 
Auto cat.— E CE CW A1 = 4.85  102 [26] 1
r ¼ A1 e RT C A C B  A1 :
2 Ke E = 14,300 cal/mole mole:s
Ke = 4 C:mol/L
r:mol/Ls
 M.A. Santaella et al. / Chemical Engineering and Processing 96 (2015) 1–13
disadvantages when using H2SO4, including the darkening of
products, equipment corrosion, required neutralization, and
infeasibility of reusing and recycling the catalyst. Despite these
disadvantages are avoided when using solid catalysts, almost none
commercial heterogeneous catalyst, including resins, has been
found to increase reaction rates to EtAc better than sulfuric acid
under the same operating conditions and acid equivalent loadings.
In Table 2, a summary of reaction rate expressions reported
using different catalysts is presented, together with the corre-
sponding parameters. Since some kinetic expressions have been
obtained including different thermodynamic models than the one
selected here (activity-based kinetics [16,17]), only the reports
including concentration-based or molar fraction-based kinetics are
listed. Expressions with no registry of experimental validation
were excluded [18]. In order to evaluate reliability of the kinetic
expressions, a comparison of these models was performed under
the same operating conditions that are feasible to reach within a
reactive distillation (RD) column. Despite rate equations were used
as published, modifications in some models were necessary to
reproduce experimental data reported on the original articles. For
instance, using mass fraction instead of mass percentage to
calculate the rate constant in the work by Atalay [19] reproduced
the exact values of the kinetic constants reported in the original
paper. In some extreme cases, reported rate expressions were
inconsistent taking into account the units of the parameters and
the variables of the model. Some comments over these rate models
are also included in Table 2.
Liquid phase esterification profiles obtained with kinetic
expressions from Table 2 are presented in Fig. 1. Comparison
was performed under batch operation using an equimolar feed of
acid and alcohol, at 80
C, and with the same amount of acid
equivalents corresponding to a loading of 0.5% wt of H2SO4.
Expressions reported at a single temperature (i.e., Amberlyst 15
[20] and hydrochloric acid [21]) were excluded from Fig. 1. As
expected, most profiles show an equilibrium conversion around
64%, and a large improvement when using catalysts compared with
non-catalytic reaction. However, the non-catalyzed reaction
0.8
0.7
0.6
0.5
HAc Conversion
0.4
0.3
0.2
0.1
0 Auto cat -2
0 200 400
Time (min)
Fig. 1. Effect of catalysts on conversion of acetic acid (HAc) during esterification with ethanol (EtOH) at 80
C and molar ratio 1:1.
5
reaches a conversion up to 20% at high residence times. This
indicates that the auto-catalytic reaction rate has to be included
when modeling the non-catalyzed reactive stages in RD columns,
mainly at high temperatures. A kinetic model reported for
Amberlyst 15 [22] seems to be inconsistent because it reaches
equilibrium condition in less than one hour, and exceeds the rate of
reaction obtained with sulfuric acid.
Another important outcome of this validation is the unexpected
behavior observed on the Alejski and Duprat’s kinetics [23] using
sulfuric acid as catalyst (H2SO4-1 in Table 2). This profile does not
show significantly faster kinetics than most resin catalyzed
reactions, which does not agree with literature reports or with
the previous experience of the authors. It is interesting to point out
that this model (apparently incorrect), has been widely used in
most studies of EtAc production via RD using sulfuric acid as
catalyst [5,12,27–29]. In comparison, the kinetic profile obtained
for H2SO4 catalyzed reaction using Atalay’s model [19] (H2SO4-2
in Table 2) does show the expected behavior, reaching equilibrium
condition much faster than heterogeneous catalysts.
In order to compare different processing alternatives for EtAc
production under the best reaction conditions, sulfuric acid was
selected as catalyst for further simulations. Taking into account the
above, simulations were performed using the kinetic model
reported by Atalay [19], and also the auto-catalytic kinetic model
[23] was included when needed. For simplification, the catalyst
loading was assumed constant along the reactive stages within the
columns.
Unlike mass or molar-fraction, the molarity-based kinetic
expressions used in this work are temperature dependent. To
ensure good kinetic predictions, it is necessary to have good
representation of the temperature dependence of densities in the
reactive system. In Figs. S1 and S2 of the supplementary material it
is presented a comparison of experimental and predicted densities
of pure components and reactive mixtures respectively at different
temperatures (Data also included in Supplementary material). As
verified, liquid density predictions are in good agreement with
Amb15-1
Amb15-2
Amb35
Amb36
Amb70
Purolite CT179
H2SO4 -1
H2SO4 -2
Auto cat -1
600 800 1000
6 M.A. Santaella et al. / Chemical Engineering and Processing 96 (2015) 1–13

experimental reports, hence, the molarity-based kinetic model can and raw materials specifications, catalysts, kinetics, and a variety of
be used with confidence. thermodynamic models to describe phase equilibrium.

3.3. Processing alternatives for ethyl acetate production
The liquid esterification of acetic acid (HAc) with ethanol
(EtOH) is a reversible reaction. Equimolar amounts of acid and
alcohol reach an equilibrium conversion of around 64% at 80
C.
Therefore, to increase conversion any of the following alternatives
may be applied: the use of excess of acid or alcohol, the reduction
of temperature as the reaction is slightly exothermic, or the use of
simultaneous reaction and separation techniques to displace
equilibrium towards product formation. The temperature reduc-
tion in this case is not recommended since the equilibrium
constant barely increases (71% conversion at 25
C, with a molar
basis equilibrium constant of 6,5), and there is a more pronounced
negative effect caused by a reduced reaction rate, increasing the
required reactor volume. Likewise, the use of an excess of reactant
will require recovery and recycle. If excess of alcohol is used,
ternary separation of EtOH–EtAc–H2O is challenging because the
azeotropic behavior of the mixture. When an excess of HAc is used,
equipment has to be corrosion-resistant and it is necessary to
accomplish HAc separation from water, which is also difficult and
generally requires entrainers in distillation. In this sense, the
preferred alternative for process intensification is to carry out
simultaneous reaction and separation, for example by using
reactive distillation (RD). Table 3 summarizes main characteristics
of some recent attempts to improve EtAc production by using RD
under different configurations (e.g., conventional, pressure swing,
multiple reactions, thermally couple distillation, dividing wall
column). As observed, these investigations use different product
3.3.1. Base case—conventional process
A scheme of the conventional continuous process to produce
EtAc is presented in Fig. 2. In this process, the esterification
reaction is carried out in a stirred tank reactor (Rx) at 80
C with a
residence time of 200 min where excess HAc is used to shift the
reaction towards the ester production. The out coming stream
from the reactor is directed to the principal distillation column
(DC) to separate EtAc from the remaining HAc. The distillate from
the principal column is cooled down and mixed with pure water to
trigger phase separation in a decanter. The organic phase from the
decanter, containing EtAc above 90% wt, is purified in a recovery
column (RC) to obtain pure EtAc in the bottoms. The distillate from
the recovery column is once again returned to the main column
(DC). Although not included in [44], an azeotropic distillation
column (AD) which uses EtAc as an entrainer is generally used to
recover unreacted HAc and to recycle it back to the reactor. The
distillate from the azeotropic column, containing mainly H2O and
EtAc is cooled down to trigger phase separation and the organic
phase is returned to the principal column. In addition to the
complexity and large energy consumption of this process, there are
raw materials and product loss within the aqueous outlet streams.
3.3.2. Reactive distillation
Various reactive distillation configurations have been studied
for the production of EtAc [5,12,14,41,42,29,45–47]. However, most
processes can be summarized in a scheme that includes a RD
column, a decanter and a purification stripper (RC), as shown in
Fig. 3. HAc inlet stream is mixed with the recovered bottoms from

Table 3
Reports on intensified processes involving reactive distillation for ethyl acetate production.

Technology Catalyst Thermodynamic Reactants purity (%) EtAc purity (wt%) Ref.
model
RD H2SO4 UNIFAC:IDEAL Pure 82,2a [27]
RD + pressure swing H2SO4 WILSON:IDEAL Pure 99,5 [16]
RD – WILSON:IDEAL Pure 98,6c [30]
RD H2SO4 KOMATSU HAc:95,2 mol EtOH:82,2 mol 83,3a [23]
BRD – Constant a – – [18]
RD Purolite UNIQUAC-HOC Pure 82,4a [31]
RD + coproduction of butanol Potassium Butanolate NRTL:IDEAL Pure 99,7a [32]
Heat integrated RD H2SO4 NRTL:IDEAL Pure 99,7a [28]
RDWC Equilibrium NRTL:IDEAL Pure 82,6a [33]
RD H2SO4 NRTL:IDEAL Pure 80,4a [34]
R and RD Purolite NRTL:IDEAL HAc:glacial 100wt 92,8 [35]
EtOH: p.a. grade 96 vol
BRD – NRTL:IDEAL Pure 97,3b [36]
RDWC Equilibrium NRTL:IDEA Pure 83,4a [37]
RD Non-cat NRTL:IDEAL HAc:49,63 mol EtOH48,08 mol H2O:2,29 mol 73,0a [38]
SBRD Purolite NRTL:IDEAL Pure 63,3b [39]
RD H2SO4 NRTL:PR:IDEAL EtOH:88,17 mol HAc:100 mol 99,6a [40]
RD H2SO4 and Purolite NRTL:HOC EtOH:87 mol HAc 95 mol 99,5a [12]
RD H2SO4 and Purolite NRTL:HOC EtOH:87 mol HAc:95 mol 99,5a [41]
RD Amberlyst 15 and Purolite NRTL:HOC Pure 99,5a [42]
RD H2SO4 NRTL:HOC ETOH:95,2 mol HAc:82,2 mol 99,5 [5]
RD + coproduction of butyl acetate Amberlyst 15 NRTL:HOC Pure 99,5 [13]
RD Amberlyst 35 NRTL:HOC EtOH:86,98 wt 99,95 [14]
HAc:96,75 wt
RD Purolite NRTL:HOC EtOH:87 mol 99,5a [43]
HAc:95 mol
RD H2SO4 NRTL:HOC EtOH:82,2 mol HAc:95,2 mol 99,5 [29]
RDWC H2SO4 (at equilibrium) NRTL:RK Pure 99,5 [6]

D: distillation, RD: reactive distillation, BRD: batch reactive distillation, SBRD: semi batch reactive distillation, RDWC:reactive distillation dividing wall, a; relative volatility.
a
Calculated from reported molar composition.
b
Calculated from molar purity assuming EtOH as other compound.
c
Calculated from transformed molar coordinate.
 M.A. Santaella et al. / Chemical Engineering and Processing 96 (2015) 1–13 7

H2O

RC

HAC

DC H2O
ETOH
Rx

ETAC
H2O

AD

Fig. 2. Conventional production process—adapted from Kirk-Othmer [44] and completed (dotted line) with water-acid azeotropic distillation column for acid recovery and
recycle.

EtAc is the heavier component, and can be obtained as a bottom

product in a subsequent recovery column (RC). The distillate of this
column is returned to atmospheric condition and recycled to the
RD column. Unlike the previous alternatives, pressure swing
scheme has only two outlet streams, assuring total conversion and
complete use of reactants. For this reason, the bottom product of
the atmospheric RD column is nearly pure water. In general, this RD
configuration would be preferable among others, as both reaction
products are removed simultaneously from the reactive media
through different outlet streams.

3.3.4. Dividing wall column with reactive reboiler

The use of a dividing wall column with a reacting reboiler
(DWC) to reduce energy consumption in the synthesis of EtAc has
been recently reported [6,33,48]. In this case, acid and alcohol are
Fig. 3. Reactive distillation process—adapted from Tang et al. [5] (shaded region fed into the reboiler of the DWC where the reaction takes place,
corresponds to catalytic section). and a side stream is withdrawn to remove water. Though high
energy savings over the conventional RD column were claimed
the RD before entering the column. The distillate of the RD column (30%), the purity of the EtAc product remained below 75% wt
is introduced to a decanter where water is added to facilitate phase since no further recovery was implemented. In addition to the loss
splitting. The organic phase is partially refluxed to the RD and
partially driven to an additional stripper (RC) for final purification
where EtAc is removed in the bottoms product. If cost-effective, the
aqueous phase produced in the decanter that contains mainly H2O
(82% wt) and roughly equal amount of alcohol and ester, is purified
to be recycled, otherwise treated before disposal. Because some
reactant and product are lost within the aqueous phase, excess
EtOH is required and EtAc recovery is reduced.

3.3.3. Pressure swing reactive distillation

As discussed in some studies [16,28], it is feasible to use a
pressure swing strategy to produce high purity EtAc running the
RD at atmospheric pressure and the recovery column at higher
pressure (350 kPa). The reported scheme is presented in Fig. 4.
While bottoms from RD are mainly water, composition of the
distillate approaches the ternary azeotrope. Using a pressure
Fig. 4. Pressure swing reactive distillation process—adapted from Seferlis and
swing, the azeotrope moves into a distillation region where pure Grieving [16]
8 M.A. Santaella et al. / Chemical Engineering and Processing 96 (2015) 1–13

Fig. 5. Dividing wall column process with reactive reboiler—adapted from

Hernandez et al. [33] (dotted area) by adding EtAc refining column. (shadowed
condenser was replaced by a two phase decanter).

of raw materials within the product, the side stream from the main
column also contained large amount of HAc, which cannot be
recycled before water removal (difficult separation as mentioned
in Section 3.3.1).
To perform a fair comparison with other technologies, the
original DWC scheme [33] was coupled to a side stripper (RC) in
order to obtain the required high purity EtAc, as presented in Fig. 5.
As in the original report, the catalyzed reaction is assumed to take
place in the reboiler only. The distillate from the DWC is directed to
a decanter where water is added to induce phase splitting. A
fraction of the organic phase is refluxed to the DWC and the
remaining fraction is directed into a side stripper for purification.
Maintaining the original concept, a side stream is withdrawn from
the main column containing over 75% wt water.

3.3.5. novel proposal—reactive dividing wall column

Taking into account the improvements reported for the
intensified schemes presented above, a novel process was
conceptually constructed. Analyzing the RD configuration pre-
sented by Tang et al. [5] and the DWC presented by Barroso-Muñoz
et al. [6], it was expected that a configuration of a reactive dividing
wall column (RDWC) like the presented in Fig. 6a, would have
resulted in higher conversions and energy savings. This was
expected because the continuous removal of water would help
reducing the remixing effect into the reactive section. This kind of
technology is far from new, and one of the first descriptions of a
RDWC was reported by Keibel in 1984 [49].
After a preliminary evaluation of a proposed RDWC scheme
described in Fig. 6a, large losses of EtAc were observed within the
aqueous stream leaving the overhead decanter (7% w/w). An
attempt to improve EtAc recovery led to the modified configuration
presented in Fig. 6b with water removal as a side stream. According
with a preliminary analysis of this design, EtAc losses were
minimized dramatically. However, there was either some HAc
losses in the side stream or an internal water recycle to the reboiler
and to the reactive stages, all resulting in lower conversions.
Finally, a new RDWC configuration presented in Fig. 6c was
proposed and the preliminary analysis indicated improved
performance compared with the previous ones. This RDWC
configuration was evaluated in further detail in this work, and it
consists of a reactive dividing wall column that uses two reboilers.
In a basic sense, the proposed scheme represents a main RD
column with a side stripper, both located inside the same shell. The
Fig. 6. Different proposed RDWC configurations. (a) Side stream to remove water.
Organic phase used as reflux. (b) Side stream to remove water. Both organic
(partially) and aqueous (totally) phases refluxed back to the column. (c) Additional
reboiler (right) and additional bottoms stream (right) to remove water. Both organic
(partially) and aqueous (totally) phases refluxed back to the column.
dividing wall goes from the last stage up to the rectifying section of
the main column. As in the other schemes, this new configuration
also requires a decanter and an additional recovery column. This
scheme can be modeled as a thermally coupled reactive distillation
sequence with side stripper, so an additional reboiler was required.
In this case the catalyst is fed and contained within the main
reactive column where the appropriate catalytic kinetics was used.
In the other side of the wall, self-catalytic reaction can occur, and
the corresponding kinetic model was included. The organic phase
withdrawn from the decanter was partially refluxed to the dividing
wall column and the remaining fraction is sent to the recovery
column (RC) for further purification. The aqueous phase from the
decanter was directed to the non-reactive section of the dividing
wall column in order to remove pure water from the bottom
 M.A. Santaella et al. / Chemical Engineering and Processing 96 (2015) 1–13 9

stream. The base case considered a number of reactive stages equal with stream exchange between them, using RadFrac model
to those for the reported RD configuration [5]. A preliminary (equilibrium stage model). In the reported DWC, gas and liquid
sensitivity analysis was carried out to determine the position of the exchange was implemented in the top and in the bottom of the
diving wall of this configuration. “prefactionator column”. In this new process, vapor and liquid
interconnection was only implemented in the top of the “pre-
3.4. Simulation and optimization fractionator”. During simulating the conventional process [44], a
water-acid azeotropic distillation column was added for acid
During simulations raw material feed streams and EtAc final recuperation. In the DWC simulation, a head decanter and a
product were specified according with technical datasheets recovery column were added to achieve the EtAc desired purity.
information [50–52]. Such specifications were selected to repre- Raw materials, products, and fuel prices were obtained from
sent characteristics of commonly processed feedstock at industrial updated reports [53–55], and were used for the calculation of TAC
facilities (Ethanol at 92.2 wt% and acetic acid at 98.5 wt%). Side (Eq. 1). Cooling water costs were considered negligible compared
reactions and byproduct formation were neglected; however when with fixed costs and heating costs, and heat integration was not
using sulfuric acid as catalyst, ethanol etherification and some considered in this study. As the catalyst is homogeneous, bubble
oxidation reactions might occur mainly at high temperatures. cap tray type columns were used in the economic evaluation. In the
Sulfuric acid was assumed to remain in the liquid phase, and specific case of the RDWC, the diameter of column was computed
therefore its concentration in all reactive stages was defined as 1% based upon the total transversal area in both sides of the dividing
wt. The esterification reaction was considered in all sections of the wall, by adding the area required in the reactive side and that of the
column: catalytic in the reactive stages and self-catalytic in the side stripper. The cost of the dividing wall was consider negligible.
non-reactive sections. A complete description of the design specifications of the
As there is no RDWC model within Aspen Plus software, these distillation columns from each configuration is listed in Table 4.
schemes where simulated using two coupled distillation modules Both continuous and discrete variables were considered during
optimization of each configuration, and the flowchart presented in
Fig. 7 describes the design-optimization algorithm. In the
Base Case flowchart “Nvars” is the number of discrete variables, “Mo” is a
vector of Nvars size with the base case values, “Ones” is a vector of
Nvars size with the value 1 in each position, “M” is a vector of Nvars
size that holds the values of the discrete values to be evaluated. The
Base Case seed values of variables used to initiate the optimization process
Simulaon were established according with the reports of the original papers
Modify
(where configurations were obtained from). The continuous
Design
variables adjusted during optimization were: molar feed of EtOH,
molar feed of HAc, either total reflux ratio (when condenser was
Base case meets No used) or organic phase reflux ratio (when decanter was used),
desired EtAC bottoms molar flow rate, liquid and vapor molar flow rate across
purity? the two sections in the dividing wall columns (when applicable),
reboiler heavy duty, water flow to decanter, decanter temperature
Yes (when applicable), distillate to feed flow rate (when applicable).
Discrete variables evaluated in the sensitivity analysis were: feed
M=Mo-Ones
stage location, column number of stages and number of stages
above and below dividing wall (when applicable).
i=1
4. Results and discussion
j=-1
After achieving convergence for all the different base-cases
using the operating conditions reported in literature, the
M(i)=Mo(i)+j computational time required to achieve an optimal solution for
each configuration was c.a 24 h (using an average laptop). Main
results for the optimized conventional and intensified processes
Minimize: TAC are presented in Table 5, and the energy savings compared with the
Method: SQP traditional process are presented in Fig. 8. According with these
Constrains: Purity>99.5wt% results heating duty is proportional to the cooling duty, so energy
costs might be calculated merely from either heating or cooling
duty as suggested in the methodology.
No As reported in literature, the RDWC and DWC show high energy
j>1? j=j+1 savings with respect to the conventional process (Fig. 2) of 47% and
39% respectively. On the other hand, it is surprising to note that the
Yes traditional RD scheme results in large columns with the highest
energy demand of all configurations (even larger than the
No conventional process). Besides, the RD scheme needs a large
i>Nvars? i=i+1 reflux ratio in the main column compared with the recovery
column where most mass flow of streams is processed. This
Yes behavior explains the poor sustainability indicators of the RD
End configuration as we shall later analyze. All other intensified
configurations have in the recovery column a higher reflux ratio
Fig. 7. Design-optimization procedure for each case study. than in the main column.
10 M.A. Santaella et al. / Chemical Engineering and Processing 96 (2015) 1–13

Table 4
Design specifications values and ranges for each configuration.

Design specification Conventional RD RD + PS DWC RDWC

(intervals*)
Column DC AD RC RD RC RD RC DWC RC RDWC RC
Reflux Ratio 0,01-4 0,01-4 0,001-10** N/A 0,001-10 0-10 0,5-1** 0,001-4 N/A
Reboiler Duty (kW) 200–2000 0–5000
Reboiler Duty right side (kW) 40–220
Material SS SS CS SS CS SS CS SS CS SS CS
Pressure (bar) 1 1 1 1 1 1 3,5 1 1 1 1
Tray type Bubble Bubble Bubble Bubble Bubble Bubble Bubble Bubble Bubble Bubble Bubble
cap cap cap cap cap cap cap cap cap cap cap
Number stages 14–16 19–21 9–11 28–30 8–10 22–24 15–17 39–41 10 28–30 9
Feed stage for reactive 13–15 14–16 4 9–11 1 5–7 7–9 Reb 1 9–11 1
mixture
Feed stage for ethanol stream 26–28 15–17 Reb 27–29
Residence time (min) 2 2 2 2 2 2 2 2 2 2 2
D/F 0,6–0,8 0,3–0,9 0,6–0,8 0,9–1 0,6–0,7 0,1–0,85 0,01–0,85 1
B (kmol/h) 0–50
B Ratio 4,374

Stages above wall 8–10 1–3

Stages below wall 10–12 0
Lateral stream stage 19–21
Lateral stream flow 100–200
L liq (kmol/h) 170–200 60–80
V vap 350–380

SS: stainless steel CS: carbon steel reb: reboiler.

*
Some design specifications are not fixed because an optimization ranges was used instead.
**
Reflux Ratio from the organic phase leaving the decanter.

the one with the lower energy consumption or smaller equipment.

Fig. 8. Energy savings over conventional process.
Fixed, variable and total costs of evaluated processes are
summarized in Table 6. As it can be seen, more than 90% of the TAC
are concentrated in the raw materials while energy costs and fixed
costs only represent c.a 5% and 1% respectively. From this
distribution it can be expected that the more cost effective
alternative to be the one with the higher productivity rather than
Therefore, RDWC, RD + PS, and RD configurations ended up with
high economic savings with respect to the conventional process,
mainly because of the enhanced conversion and the consequent
higher productivity promoted by the continuous displacement of
the equilibrium reaction and secondly due to the lower energy
consumption in the separations stages. Surprisingly and although
the DWC scheme achieves the high energy savings mentioned
above, the use of this processing alternative increases total costs
mainly due to the lower conversion. Raw material losses in the side
stream are the major cause of such inefficient performance. To
maintain the specified production rate under DWC configuration,
large feedstock consumption and large processing units are
required. As we shall analyze below, material losses do not only
affect the TAC, they also have a radical effect over many
sustainability indicators.
The complete set of results from the optimized processes can be
consulted in Figs. S3–S7 in the supplementary material. The
sustainability indicators obtained under these optimal operating
conditions for the different configurations are summarized in
Table 7.
As reported before, energy consumption in most intensified
processes is substantially reduced, except in the case of the RD
configuration. This configuration requires a large reflux ratio to
achieve high conversions. As the acid inlet is located at the top of

Table 5
Main characteristics of the optimized reactive distillation processes.

Process EtOH HAC Heating Cooling Number of Number of Reflux ratio Reflux ratio Fresh water D (m) D (m)
stream stream duty duty stages main reactive main column recovery inlet (kg/h) main recovery
(kg/h) (kg/h) (kW) (kW) column stages column column column
Conventional 10219 10079 26582 26664 15 1 0.93 0.67 19309 4,3 3,1
RD 7465 8680 29213 29704 29 18 5,14a 1,87 1247 7,4 2,7
RD+PS 7397 8601 19302 19203 24 18 0.25 1.5 – 4,1 3,4
DWC 1357 15936 16217 16565 40 1 0,78 1.75 29575 4,3 3,1
RDWC 6989 8541 14098 14155 29 19 0,03 2,08 1516 4,5 2,4
a
Reflux Ratio from the organic phase leaving the decanter.
 M.A. Santaella et al. / Chemical Engineering and Processing 96 (2015) 1–13 11

Table 6
Fixed, variable and total annual costs.

Process Energy costs MM.USD/yr Raw Materials costs MM.USD/yr Variable costs MM.USD/yr Fixed costs MM.USD/yr TAC MM.USD/yr
Conventional 6,5 123,5 130 2,3 132,3
RD 7,1 97,3 104,4 0,8 105,2
RD + PS 4,7 96,5 101,1 0,5 101,6
DWC 3,9 177,5 181,4 0,6 182
RDWC 3,4 93,4 96,8 0,5 97,3

Table 7
Sustainability indicators of optimized reactive distillation processes for EtAc production.

Conversion Recovery Productivity MI EW E MP EI

Conventional 0.98 0.86 0.84 1.58 2.23 0.34 0.63 2.17
RD 1 0.97 0.97 1.26 0.42 0.05 0.79 2.38
RD + PS 0.98 1 0.98 1.25 0.31 0.04 0.8 1.69
DWC 0.71 0.75 0.53 2.3 3.82 1.02 0.44 1.32
RDWC 0.99 1 0.99 1.21 0.39 0.01 0.83 1.15

the reactive zone of the column, some acid is pulled to the top and
lost within the distillate. To avoid such losses, and similarly to the
AD column of the conventional configuration (Fig. 2), a high reflux
ratio of the organic phase is needed. With this, the EtAc rich stream
acts as an extracting agent that helps pushing the acid back to the
reactive zone improving conversion. This large reflux explains why
RD presents the worst performance according with the EI indicator,
being even 10% more energy intensive than the conventional
process. Comparatively, and despite the fact that a higher pressure
is required, the RD+PS configuration requires a lower reflux ratio
than the RD process (i.e., less energy), since the recycle stream
from the high pressure column helps keeping the acid content low
in the condenser from the reactive column. The herein proposed
RDWC configuration offers the lowest energy intensity and the
higher savings from all intensified processes since it avoids
remixing during multicomponent distillation, and also allows
selective product separation during esterification inside the same
unit. In this case, the low organic phase reflux flow rate (see
Table 5) is possible due to the recycle of the aqueous stream from
the decanter back to the column, since it lowers the top stage
temperature and increases the actual amount of liquid refluxed to
the column. In Fig. 9, the composition profiles of the most cost-
effective configuration (RDWC) are presented. While the most of
the HAc is maintained in the reactive side, EtOH is separated from
water in the side column and sent to the reactive stages assuring
complete conversion. Almost pure water is removed from bottoms
of the side stripper and the azeotrope is send to EtAc purification
column. In the modeling specifications, it was established that
both the reactive and side columns had the same number of stages.
However, as observed in the composition profiles of Fig. 9, the side
stripper is oversized. In this case, the use of different column
internals or a lower number of trays in the side column might
decrease even more the capital costs of the RDWC. It is interesting
to notice how the intensified configurations (RD, RD + PS and
RDWC) have similar conversion to that of the conventional scheme
but different productivity. The overall high conversion from the
conventional process is achieved by recycling almost all of the
limiting reagent of the process, which is the HAc from the AD
column. However, the low recovery capacity of the decanters in the
conventional scheme, where a large amount of fresh water is
needed to trigger phase separation explain the low productivity.
On the other hand, the productivity of DWC scheme is very low

1
1 2
2 3
3 4
4
5
5
6 6
7 7
8 8
9 9
10 10
11 11
Stage Number

12
Stage Number

12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
20 Catalytic stages 20
21 21
22 22
23 23
24 24
25 25
26 26
27 27
28 28
29 29
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1
Mole Fraction Mole Fraction
Fig. 9. Liquid phase composition profiles in a reactive dividing wall column configuration for ethyl acetate production using sulfuric acid as catalyst. Left—main reactive
column, right—side stripper. (—) HAc, (--) EtOH, (—) EtAc, ( . . . ) H2O.
12 M.A. Santaella et al. / Chemical Engineering and Processing 96 (2015) 1–13
mainly because of its low conversion. Which can be explained from
the DWC configuration, where the limiting reagent (HAc)
withdrawn within the side stream is never recycled back to the
reactive reboiler.
Regarding the MI factor for the different technologies evaluated
in this work, differences from RD, RD + PS and RDWC, can be
explained by the different excess reagent amounts that are fed to
each system. While RD and RD+PS required a 5% excess of ethanol,
the RDWC optimum solution uses nearly equimolar feed, which
combined with high productivity, makes mass utilization (MI) very
efficient. With respect to the E-Factor and E-W-Factor, DWC
produces more waste than the conventional process and other
intensified alternatives once again because of the unreacted HAc
leaving the process form the side stream, which in this case is
assumed as waste. Furthermore, the E-W factor shows that the
RDWC scheme generates 5 times less waste than other intensified
and profitable configurations (RD and RD + PS) and 40 times less
waste than the conventional scheme. In this case the RDWC
produces extremely low waste amounts because this novel design
does not allow unreacted HAc nor EtOH to leave the process, as
well as the fact that there is no EtAc lost within the waste water
stream.
Although the RD + PS configuration has low E-factor because no
additional water is needed, the RDWC shows a better performance
because it uses the water obtained as product of the side column to
trigger phase separation in the decanter. This, together with the
lower TAC, low energy and materials intensity, makes RDWC
alternative to be the more attractive from an economical and
sustainable point of view.
5. Conclusions
An assessment of different intensified processes for the
production of ethyl acetate through Fisher esterification has been
carried out. The analysis allowed to select reliable thermodynamic
models to describe the multicomponent phase equilibria involved
in the process, as well as a homogeneous kinetic model that
correctly describes the reaction rate using sulfuric acid as catalyst.
The thermodynamic and kinetic models selected together with
commercial specifications of raw materials and products were
implemented for the simulation of different intensified processes
proposed in the literature. Conventional process, reactive distilla-
tion, reactive distillation with pressure swing, dividing wall
column with reactive reboiler, and a novel process consisting of
a reactive dividing wall column, configurations were considered.
The optimization of those configurations was carried out by
minimizing total annual costs, using sensitivity analysis and SQP
approach. Finally, conversion, productivity, mass intensity, mass
productivity, E-Factor, EW-Factor, and energy intensity were
selected as sustainability indicators of the technologies evaluated.
According with results, reactive distillation offers lower costs and
better sustainability although having higher energy intensity than
the conventional process. DWC although less energy intensive, has
the higher costs and the worst sustainability indicators. Reactive
distillation with pressure swing offers energy savings, lower costs
and better sustainability than the conventional process. Amongst
all the evaluated alternatives, the novel reactive dividing wall
column configuration developed here offers the highest energy
savings, the lower annual costs, and the overall better sustainabil-
ity indicators.
Acknowledgement
This work is supported by Colombian Administrative Depart-
ment of Science, Technology and Innovation (COLCIENCIAS) under
the project: “Producción de plastificantes a partir de acido cítrico
usando procesos híbridos de reacción y separación simultanea”
cod. 1101-569-33201.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.cep.2015.07.027.
References
[1] J.P.M. Sanders, J.H. Clark, G.J. Harmsen, H.J. Heeres, J.J. Heijnen, S.R.a. Kersten,
et al., Process intensification in the future production of base chemicals from
biomass, Chem. Eng. Process. Process Intensif. 51 (2012) 117–136, doi:http://
dx.doi.org/10.1016/j.cep.2011.08.007.
[2] R. Srinivasan, N.T. Nhan, A statistical approach for evaluating inherent benign-
ness of chemical process routes in early design stages, Process Saf. Environ.
Prot. 86 (2008) 163–174, doi:http://dx.doi.org/10.1016/j.psep.2007.10.011.
[3] N.T.H. Thuy, Y. Kikuchi, H. Sugiyama, M. Noda, M. Hirao, Techno-economic and
environmental assessment of bioethanol-based chemical process: a case study
on ethyl acetate, Environ. Prog. Sustain. Energy 30 (2011) 675–684, doi:http://
dx.doi.org/10.1002/ep.
[4] T. Van Gerven, A. Stankiewicz, Structure, energy, synergy, timesthe
fundamentals of process intensification, Ind. Eng. Chem. (2009) 2465–2474.
[5] Y. Tang, H.-P. Huang, I.-L. Chien, Design of a complete ethyl acetate reactive
distillation system, J. Chem. Eng. Jpn. 36 (2003) 1352–1363.
[6] F.O. Barroso-Muñoz, S. Hernández, J.G. Segovia-Hernández, H. Hernandez-
Escoto, A.F. Aguilera-Alvarado, Thermally coupled distillation systems: study
of an energy-efficient reactive case, Chem. Biochem. Eng. Q. 21 (2007)
115–120.
[7] L. Kelley, F. Mirasol, Ethyl acetate, ICIS Chem. Bus. 281 (2012) 43–44.
[8] J.M. Douglas, Douglas—Conceptual Design Of Chemical Processes, McGraw-
Hill, 1988.
[9] R.a Sheldon, Atom utilisation, E factors and the catalytic solution, Comptes
Rendus l’Académie Des. Sci. - Ser. IIC - Chem. 3 (2000) 541–551, doi:http://dx.
doi.org/10.1016/S1387-1609(00)01174-9.
[10] C. Jimenez-Gonzalez, D.J.C. Constable, Green Chemistry and Engineering: A
Practical Design Approach, Wiley, New Jersey, 2011.
[11] J. Gmehling, J. Menke, J. Krafczyk, K. Fischer, Azeotropic Data, 2nd ed., Wiley-
VCH, Weinheim, 2004.
[12] H. Huang, I. Chien, H. Lee, Plantwide Control of a Reactive Distillation Process,
in: G.P. Rangaiah, V. Kariwala (Eds.), Plantwide Control Recent Dev. Appl, First
John Wiley & Sons Inc., Taipei, 2012, pp. 319–338.
[13] H. Tian, H. Zheng, Z. Huang, T. Qiu, Y. Wu, Novel procedure for coproduction of
ethyl acetate and n-butyl acetate by reactive distillation, Ind. Eng. Chem. Res.
51 (2012) 5535–5541, doi:http://dx.doi.org/10.1021/ie202154x.
[14] I.-K. Lai, Y.-C. Liu, C.-C. Yu, M.-J. Lee, H.-P. Huang, Production of high-purity
ethyl acetate using reactive distillation: experimental and start-up procedure,
Chem. Eng. Process. Process Intensif. 47 (2008) 1831–1843, doi:http://dx.doi.
org/10.1016/j.cep.2007.10.008.
[15] W. Riemenschneider, H.M. Bolt, Esters, Organic, Ullmann’s Encycl. Ind. Chem.
(2005) 8676–8694, doi:http://dx.doi.org/10.1002/14356007.a09.
[16] P. Seferlis, J. Grievink, Optimal design and sensitivity analysis of reactive
distillation units using collocation models, Ind. Eng. Chem. Res. 40 (2001)
1673–1685, doi:http://dx.doi.org/10.1021/ie0005093.
[17] A. Orjuela, A.J. Yanez, A. Santhanakrishnan, C.T. Lira, D.J. Miller, Kinetics of
mixed succinic acid/acetic acid esterification with Amberlyst 70 ion exchange
resin as catalyst, Chem. Eng. J. 188 (2012) 98–107, doi:http://dx.doi.org/
10.1016/j.cej.2012.01.103.
[18] L.S. Balasubramhanya, F.J. Doyle III, Nonlinear model-based control of a batch
reactive distillation column, J. Process Control 10 (2000) 209–218, doi:http://
dx.doi.org/10.1016/S0959-1524(99)00024-4.
[19] F.S. Atalay, Kinetics of the esterification reaction between ethanol and acetic
acid, Dev. Chem. Eng. Miner. Process 2 (1994) 181–184, doi:http://dx.doi.org/
10.1002/apj.5500020210.
[20] K. Tanaka, R. Yoshikawa, C. Ying, H. Kita, K. Okamoto, Application of zeolite
membranes to esterification reactions, Catal. Today 67 (2001) 121–125, doi:
http://dx.doi.org/10.1016/S0920-5861(01)00271-1.
[21] J.M. Smith, Chemical Engineering Kinetics, McGraw-Hill, 1956.
[22] S.I. Kirbaslar, Z.B. Baykal, U. Dramur, Esterification of acetic acid with ethanol
catalysed by an acidic ion-exchange resin, Turk. J. Eng. Environ. Sci. 25 (2001)
569–577.
[23] K. Alejski, F. Duprat, Dynamic simulation of the multicomponent reactive
distillation, Chem. Eng. Sci. 51 (1996) 4225–4237.
[24] A. Orjuela, A.J. Yanez, C.T. Lira, D.J. Miller, Ethyl Acetate Esterification Kinetics
Using Amberslyst 15 as Catalyst, Internal Research Report, Michigan State
University, 2014.
[25] G. Hangx, G., Kwant, H., Maessen, P., Markusse, I., Urseanu, P. Moritz, et al.
REACTION KINETICS OF THE ESTERIFICATION OF ETHANOL AND ACETIC ACID
TOWARDS ETHYL ACETATE Name of Partner: DSM Authors: Intelligent Column
Internals for Reactive Separations (INTINT), (2001) 1–5.
[26] H.J. Arnikar, T.S. Dao, A.A. Bodhe, A gas chromatographic study of the kinetics
of the uncatalysed esterification of acetic acid by ethanol, J. Chromatogr. A 47
(1970) 265–268.
 M.A. Santaella et al. / Chemical Engineering and Processing 96 (2015) 1–13
[27] P.V.S.R. Chandra, C. Venkateswarlu, Multistep model predictive control of ethyl
acetate reactive distillation column, Indian J. Chem. Technol. 14 (2007)
333–340.
[28] S.V. Mali, A.K. Jana, A partially heat integrated reactive distillation: feasibility
and analysis, Sep. Purif. Technol. 70 (2009) 136–139, doi:http://dx.doi.org/
10.1016/j.seppur.2009.09.006.
[29] H.-Y. Lee, H.-P. Huang, I.-L. Chien, Control of reactive distillation process for
production of ethyl acetate, J. Process Control 17 (2007) 363–377, doi:http://
dx.doi.org/10.1016/j.jprocont.2006.10.002.
[30] A. Avami, W. Marquardt, Y. Saboohi, K. Kraemer, Shortcut design of reactive
distillation columns, Chem. Eng. Sci. 71 (2012) 166–177, doi:http://dx.doi.org/
10.1016/j.ces.2011.12.021.
[31] M. Klöker, E.Y. Kenig, A. Hoffmann, P. Kreis, A. Górak, Rate-based modelling and
simulation of reactive separations in gas/vapour-liquid systems, Chem. Eng.
Process. Process Intensif. 44 (2005) 617–629, doi:http://dx.doi.org/10.1016/j.
cep.2003.12.011.
[32] S.-J. Wang, H.-P. Huang, C.-C. Yu, Plantwide design of transesterification
reactive distillation to co-generate ethyl acetate and n-butanol, Ind. Eng.
Chem. Res. 49 (2010) 750–760, doi:http://dx.doi.org/10.1021/ie901413c.
[33] S. Hernández, R. Sandoval-Vergara, F.O. Barroso-Muñoz, R. Murrieta-Dueñas,
H. Hernández-Escoto, J.G. Segovia-Hernández, et al., Reactive dividing wall
distillation columns: simulation and implementation in a pilot plant, Chem.
Eng. Process. Process Intensif. 48 (2009) 250–258, doi:http://dx.doi.org/
10.1016/j.cep.2008.03.015.
[34] E.Y. Kenig, H. Bader, A. Gã, Investigation of ethyl acetate reactive distillation
process, Chem. Eng. Sci. 56 (2001) 6185–6193.
[35] Q. Smejkal, J. Kolena, J. Hanika, Ethyl acetate synthesis by coupling of fixed-bed
reactor and reactive distillation column—process integration aspects, Chem.
Eng. J. 154 (2009) 236–240, doi:http://dx.doi.org/10.1016/j.cej.2009.04.022.
[36] E. Re, M. Meyer, A generic feasibility study of batch reactive distillation in
hybrid configurations, AIChE J. 55 (2009) 1185–1199, doi:http://dx.doi.org/
10.1002/aic.
[37] R. Delgado-Delgado, S. Hernández, F.O. Barroso-Muñoz, J.G. Segovia-
Hernández, A.J. Castro-Montoya, From simulation studies to experimental
tests in a reactive dividing wall distillation column, Chem. Eng. Res. Des. 90
(2012) 855–862, doi:http://dx.doi.org/10.1016/j.cherd.2011.10.019.
[38] H. Bock, M. Jimoh, G. Wozny, Analysis of reactive distillation using the
esterification of acetic acid as an example, Chem. Eng. Technol. 20 (1997) 182–
191, doi:http://dx.doi.org/10.1002/ceat.270200305.
[39] Z.M. Shakor, K.A. Sukkar, Dynamic simulation of semi-batch catalytic
distillation used for esterfication reaction, Eng. Technol. (2008) 26.
13
[40] Y. Tavan, S.H. Hosseini, Design and simulation of a reactive distillation process
to produce high-purity ethyl acetate, J. Taiwan Inst. Chem. Eng. 44 (2013) 577–
585, doi:http://dx.doi.org/10.1016/j.jtice.2012.12.023.
[41] H. Lee, H., Huang, I. Chien, Design and control of homogeneous and
heterogeneous reactive distillation for ethyl acetate process, 16th Eur. Symp.
Comput. Aided Process Eng. 9th Int. Symp. Process Syst. Eng. (2006) 1045–
1050.16th European Symposium on Computer Aided Process Engineering and
9th International Symposium on Process Systems Engineering.
[42] Y.-T. Tang, Y.-W. Chen, H.-P. Huang, C.-C. Yu, S.-B. Hung, M.-J. Lee, Design of
reactive distillations for acetic acid esterification, AIChE J. 51 (2005) 1683–
1699, doi:http://dx.doi.org/10.1002/aic.10519.
[43] I.-K. Lai, S.-B. Hung, W.-J. Hung, C.-C. Yu, M.-J. Lee, H.-P. Huang, Design and
control of reactive distillation for ethyl and isopropyl acetates production with
azeotropic feeds, Chem. Eng. Sci. 62 (2007) 878–898, doi:http://dx.doi.org/
10.1016/j.ces.2006.10.019.
[44] Kirk-Othmer, Encyclopedia of Chemical Technology, 8 (n.d.) 331–338.
[45] H. Zheng, H. Tian, W. Zou, Z. Huang, X. Wang, T. Qiu, et al., Residue curve maps
of ethyl acetate synthesis reaction, J. Cent. South Univ. 20 (2013) 50–55, doi:
http://dx.doi.org/10.1007/s11771-013-1458-2.
[46] S. Hung, M. Lee, Y. Tang, Y. Chen, I. Lai, W. Hung, et al., Control of different
reactive distillation, AIChE J 52 (2006) 1423–1440, doi:http://dx.doi.org/
10.1002/aic.
[47] S.-J. Wang, D.S.H. Wong, S.-W. Yu, Design and control of transesterification
reactive distillation with thermal coupling, Comput. Chem. Eng. 32 (2008)
3030–3037, doi:http://dx.doi.org/10.1016/j.compchemeng.2008.04.001.
[48] F.O. Barroso-muñoz, S. Hernández, B. Ogunnaike, Analysis of Design and
Control of Reactive Thermally Coupled Distillation Sequences, 17th Eur, Symp.
Comput. Aided Process Eng. - ESCAPE 17 (2007) 1–6.
[49] G. Kaibel, Method of carrying out chemical reactions and for the simultaneous
fractionation of a mixture into several fractions by a distillation column,
EP126288 A2, 1984.
[50] Eastman Chemical Company, Eastman Ethyl Acetate, Urethane Grade Product
Data Sheet, Eastman Chemical Company, 2001.
[51] Eastman Chemical Company, Eastman Dilute Acetic Acid, 95% Product Data
Sheet, Eastman Chemical Company, 2000.
[52] Recochem Inc, Ethanol 95 SG (SVR 95%), Recochem Inc, 2013. http://www.
recochem.com.au/index.php/products/industrial_products/ethanols/item/
ethanol_95_sg_svr_95
[53] F. Mirasol, Ethanol, ICIS Chem. Bus. 280 (2011) 71.
[54] E. Burridge, Chemical Profile: Acetic acid, ICIS Chem. Bus. May 31- Ju (2010) 36.
[55] G.D. Ulrich, P.T. Vasudevan, How to estimate utility costs, Chem. Eng. 6 (2006)
66–69.