Publications Aashish Gaurav

Article
pubs.acs.org/EF
Transesterication of Triglyceride to Fatty Acid Alkyl Esters

(Biodiesel): Comparison of Utility Requirements and
Capital Costs between Reaction Separation and
Catalytic Distillation Congurations
Aashish Gaurav, Mateus Lenz Leite, Flora T. T. Ng,* and Garry L. Rempel
Department of Chemical Engineering, University of Waterloo, Waterloo, Ontario, N2L 3G1 Canada
ABSTRACT: The ecacy of catalytic distillation (CD) and the economic merits that it would bring into the biodiesel
production process is studied. Process ow sheets depicting conventional and CD technology are modeled in Aspen Plus, and
detailed operating conditions and equipment designs are provided for each process. The feedstock considered is soybean oil, and
the transesterication reaction for the triglyceride is considered for biodiesel production. After optimal design of both process
ow sheets to produce 10 million gallons of biodiesel per year, adhering to ASTM purity standards, a detailed cost analysis is
carried out using the Aspen economic analyzer tool to predict capital, operating, and utility costs and to calculate the cost of
production per gallon of biodiesel. Results depict CD to be a promising candidate to replicate the conversion and product purity
of conventional biodiesel processes while having signicant savings in capital (41.42% cheaper than the conventional process)
and utility (18.12% less than the conventional process) costs, thereby making it a very competitive alternative. The total
operating costs and price of production per gallon of biodiesel are only meagerly cheaper for a CD process because the most
signicant factor to the biodiesel production process is the raw material cost. For both processes, the price of production per
gallon of biodiesel after accounting for revenue generated from glycerol product is predicted to be around 1.7 dollars/gallon. The
Aspen model is exible to accommodate higher ow rates for scale-up of operations, add or remove stages of operation into the
biodiesel process, modify feedstock and stream prices, and predict associated capital and production costs.
1. INTRODUCTION
The challenge to meet the worlds growing energy demands in a
sustainable manner is a key global priority, necessitating intensive
research and development of advanced renewable energy systems
for oil-based energy and other fossil fuels. Energy security,
environmental concerns, foreign exchange savings, and socioeconomic issues have made clean biofuels derived from biomass or
biological sources a hot topic. Among the various alternative fuel
options, biodiesel has spurned much interest and popularity.
Chemically, biodiesel is a mixture of fatty acid alkyl esters
(FAAEs), derived from triglyceride molecules. Triglycerides and
alcohol are converted to alkyl esters (biodiesel) via a catalyzed
transesterication reaction, as illustrated in Figure 1. Glycerol
attractive as a renewable and environmentally friendly fuel option

for domestic engines; apart from being nontoxic and
biodegradable, biodiesel results in lesser air pollutants per net
energy than diesel [reduction of greenhouse gases (GHGs) by
41%2] and an estimated 93% more energy than the energy
invested in producing it.2 Biodiesel also has a higher ash point
temperature of 150 C, making its handling, use, and transport
safer than petroleum diesel and better lubricating properties that
minimize engine wear and tear. A promising alternative to
petroleum, biodiesel use and production is hence expected to rise
signicantly in the near future.
The objective of this research is to investigate the possibility of
intensifying the conventional reactor separation technology for
continuous production of biodiesel via catalytic distillation
(CD). CD is a hybrid reactor technology that integrates
heterogeneous catalytic reaction within a distillation column by
immobilizing solid catalyst particles within discrete reactive
sections.3 The distinguishing feature of CD is its ability to
simultaneously carry out the chemical reaction and product
purication within a single-stage operation, potentially resulting
in signicant energy savings as well as a reduction in operating
and capital expenditures because of process intensication. The
continuous removal of product from the reactive section via the
distillation action can lead to an increased product yield and
increased productivity, particularly for reactions that are
Figure 1. Transesterication (methanolysis) of triglyceride molecule to

FAAEs (biodiesel).
and FAAEs are then separated and puried. Rapeseed and

soybean oils are the leading vegetable oils for biodiesel
production. The conventional catalysts for the transesterication
reaction of these vegetable oils are predominantly basic mostly
alkaline or alkaline earth oxides or alkoxides.1 Biodiesel is very
2013 American Chemical Society
Received: September 4, 2013

Revised: October 9, 2013
Published: October 10, 2013
6847
dx.doi.org/10.1021/ef401790n | Energy Fuels 2013, 27, 68476857
Energy & Fuels
Article
production process, especially in terms of reduction of water

usage for cleaning the products because water is now considered
as a valuable resource. In addition, homogeneous catalysts carry
the disadvantage of causing severe environmental hazards in the
case of spillage. Another novelty of our research is use of single
software (Aspen) comprehensively for both simulation and
equipment sizing and cost analysis. Most past researchers have
carried out economic analysis of their simulations by rst sizing
the equipment through non-iterative calculations and then
employing published equipment cost correlations9 or employing
another software (for example, Chemsoft) to estimate capital
costs using the design data.6 Aspen Plus is one of most
extensively used chemical engineering packages, and use of a
single, consistent software environment for simulation, equipment sizing, and cost analysis reduces estimation variation and
gives more accurate and reliable costing results.
To highlight the merits of CD, two process congurations
depicting continuous production of biodiesel are modeled in Aspen
Plus. Conguration A (Figure 2) is a traditional reactor separation
owsheet, where transesterication takes place in the reactor and the
products are separated in a column. Conguration B (Figure 3) is a
CD column that incorporates both the reaction and separation in one
column. For the same biodiesel production capacity, a comparison of
the total energy requirements and the capital costs is then performed
between the two process congurations. The savings in energy
requirements and capital costs are then quantied to relate the
eectiveness of CD compared to the conventional process.
equilibrium-limited. Other potential advantages of CD include

the mitigation of catalyst hot spots, improved temperature
control, and enhanced energy integration because of conduction
of an exothermic chemical reaction in a boiling medium with
in situ separation via distillation. The transesterication of
triglyceride meets the design criteria of CD because the products
glycerol and FAAEs have a signicant volatility dierence that
makes separation by distillation favorable (properties listed in
Table 1). Second, the exothermicity of the reaction as reported in
Table 1. Boiling Point and Density of the Components at 1
atm and 25 C
molecular weight (g/mol)
boiling point (C)
density (g/cm3)
triolein
methanol
methyl oleate
glycerol
885.43
846.5
0.91
32.04
64.7
0.79
296.49
349
0.87
92.09
287.71
1.26
the literature4 and also conrmed via Aspen calculations in

section 2.1 favors CD because the energy liberated can be
eciently converted in situ to drive the distillation process and
enhance energy integration. Third, the transesterication is a
reversible reaction, and the constant removal of products should
shift the equilibrium toward the products;5 hence, CD becomes a
very encouraging option.
To the knowledge of the authors, in the literature to date, no
systematic, comprehensive work has been performed comparing
the capital and operating costs of biodiesel production using CD
to the traditional process while meeting the optimal process
conditions and purity requirements. A number of authors610
have analyzed the cost of the biodiesel production process
considering traditional reaction and separation technology and
homogeneous catalysts. None of these simulations consider use
of heterogeneous catalyst systems (that would save catalystwashing operations) or examine the suitability of CD as an
alternative to the traditional process. We feel that, despite the
slower reaction rate compared to homogeneous catalysts,
heterogeneous catalysts could bring merits to the biodiesel
2. PROCESS DESCRIPTION
Batch experiments on transesterication of soybean oil with methanol were
recently carried out in our laboratory over calcium oxide supported on Al2O3
as solid base catalysts (heterogeneous catalyst system).11 The reaction was
assumed to be pseudo-rst-order with respect to the triglyceride (soybean oil
molecule) in excess of methanol.12 Average overall reaction constants at
dierent temperatures were calculated, and the activation energy was
estimated to be 30 kJ/mol based on the Arrhenius equation.
ln k =
Ea
+C
RT
(1)
Figure 2. Conguration A: traditional reactor separation ow sheet for biodiesel production.

6848
Energy & Fuels
Article
Figure 3. Conguration B: CD ow sheet.
Figure 4. Conventional ow sheet for biodiesel production from biomass feedstock.

conversion per unit of volume, and they also require lower maintenance and
shutdown times.14
A schematic diagram for a conventional biodiesel production process
is shown in Figure 4. Most researchers6,8,10 studying the biodiesel
production process using homogeneous catalyst systems base their
simulations on the shown schematic (Figure 4). The basic steps are the
The current study models the above reaction kinetics in a continuous reactor.
The transesterication reaction for biodiesel production in a continuous
process can be carried out in dierent reactors, such as a plug-ow reactor or
combined stirred-tank reactor;8,13 accordingly, the reactor conditions, such as
volume and residence time, to achieve the same reactant conversion could
dier. Plug-ow and packed-bed reactors are known for achieving the highest
6849
Energy & Fuels
Article
transesterication reaction followed by alcohol recovery in a distillation

column and glycerol catalyst extraction with water from the oil phase in a
washing column. Some processes also require neutralization of the
catalyst. Separate distillation columns are used for fatty acid methyl ester
(FAME) and glycerol purication. In our ow sheet for the conventional
reactor separation conguration with a heterogeneous catalyst, no
washing column is required for extraction of the catalyst from the oil/
biodiesel phase; we are considering a plug-ow packed-bed reactor with
a catalyst lifetime of 1 year. Economics for the heterogeneous catalyst
system is discussed in section 3. We have also added a ash distillation
unit after the high-temperature, high-pressure reactor operation because
it brings enormous reduction (4 times less) in utility usage for methanol
separation.
The process simulations of the complete ow sheets of two
congurations (conventional reactor separation and CD) are carried
out in the process modeling software Aspen Plus. Although there ought
to be dierences between process simulation results and actual process
operations, simulation software tools predict reliable and pretty accurate
information on process operation because of their comprehensive
thermodynamic packages, rich component data, and astute calculation
techniques.
Process simulations generally involve dening chemical components,
the choice of an appropriate thermodynamic model, determination of
operating conditions (temperature, pressure, concentrations, etc.), and
sizing the operation (production capacity). The composition of soybean
oil is reported to be 6.8% palmitic acid, 33.4% oleic acid, 51.5% linoleic
acid, and 2.3% linolenic acid.15,16 Because the major components, oleic
acid, linoleic acid, and linolenic acid, are all composed of 18 carbon
atoms, triolein (C57H104O6) was chosen to represent the triglyceride
molecule in soybean oil. Hence, the input components are triolein and
methanol as the reactants and glycerol and methyl oleate or FAME as
the products. Figure 5 shows the reaction. All of the components, except
length, diameter, and number of tubes, we also kept a watch on the

residence time for the reaction while working in the specied pressure and
temperature ranges. We aimed at more than 95% conversion of triolein to
biodiesel to ensure ease of separation and product purity. The reaction
time for the transesterication kinetic data used in our Aspen models was
reported in the range of 812 h.11 The residence time for our Aspen
simulation for conguration A is 16 h. The reactor length aects the
conversion very strongly, and diameter and number of tubes aect the
residence time.
The heat of reaction was calculated via Aspen by running a separate
RStoic reactor model at constant temperature and pressure and
computing the reaction heat. The heat of reaction was found to be
100.17 kJ/mol per mole of biodiesel produced at 25 C and 1 atm. The
reaction is therefore exothermic, making of it a good candidate for a CD
operation. The heat of reaction calculations was not used anywhere as an
input in the simulation but was calculated to provide an understanding of
the temperature proles, process results, and energy requirements better.
The main processing equipment units for the ow sheet are shown in
Figure 6. The ow sheet elements are a mixer, a plug-ow xed-bed
reactor (PFR), a ash separator, a decanter, and two distillation columns
(RADFRAC). The mixer functions to enable the recycle of the overhead
methanol streams from the ash and distillation columns back into the
reactor. The ash separates methanol from glycerol and methyl oleate.
The decanter splits methyl oleate (biodiesel product) from glycerol. The
purpose of the rst distillation RADFRAC column (FAME purication,
DISTL1) is to purify the biodiesel from methanol, while the second
column (glycerol purication, DISTL2) separates methanol from
glycerol.
A constant feed with a molar ow rate of 100 lb mol h1 in a
methanol/oil optimum molar ratio of 9:1 (for maximum conversion11)
is fed to the reactor. The reactor conguration is shown in Table 3. A
conversion of 99.83% is achieved. The output stream from the reactor is
at 160 C and 15 atm.
Pure component physical properties obtained from the Aspen library
suggest that the product separation should be easy because they have
appreciable boiling point and density dierence (Table 1). Because the
boiling point of methanol (64.7 C at 1 atm) is signicantly lower than
that of methyl oleate (343.85 C at 1 atm) and glycerol (287.85 C at
1 atm), a ash separator was used to isolate methanol from glycerol and
methyl oleate. The ash drum works on the principle of a high-pressure
gradient to achieve a good separation. The steps were methanol
separation followed by the decantation of glycerol and nal distillation of
the products. The ash drum nal operating conditions were 125 C and
1 atm. The top stream from the ash has a methanol purity of 99%,
which was recycled back to the reactor via the mixer. Design and tear
stream specications were specied in Aspen to estimate the fresh feed
composition to the mixer to keep the methanol/triolein ratio of 9:1 at
the mixer output. The bottom stream from the ash separation unit was
taken to a decanter for separation of glycerol and methyl oleate via
gravity separation.
The decanter functions to purify the biodiesel product (methyl
oleate) from methanol and glycerol based on the density dierence and
intermolecular interaction between them. Separation using a gravity
settler has also been proposed.17 The decanter operates at 125 C and 1
atm. The lighter stream from the decanter is composed of 90.89 mol %
methyl oleate (biodiesel) and 8.77 mol % methanol. The denser stream
is composed of a glycerol mole fraction (85.32 mol %) and methanol
(14.66 mol %).
Two RADFRAC distillation columns (FAME purication and
glycerol purication) are employed for methanol recovery as well as
purication of biodiesel and glycerol. The FAME purier had ve stages
(condenser, three trays, and reboiler); the feed was fed to stage 3. The
glycerol purier only had three stages (condenser, tray, and reboiler),
with feed on stage 2. Running simulations at varying reux ratios
resulted in minimum energy requirements at 0.5; therefore, this was set
as the optimum reux ratio for both columns. The column stage
eciency was assumed to be 1, and no pressure drop was considered to
take place in the columns. The distillate was totally condensed.
The FAME purier distillation column achieved a separation of
99.99% mole fraction of methanol on the top stream (99.99% mass fraction)
Figure 5. Transesterication of the triolein molecule to methyl oleate

(biodiesel).
for triolein, had their property available in the Aspen pure component
database. The triolein molecule le was imported from the National
Institute of Standards and Technology (NIST) chemistry web book
database, and its properties were downloaded using the NIST thermo
data engine. All missing pure component properties and binary
interaction parameters were estimated by Aspen. The process-type
properties were chosen on the basis of the process. For both owsheets,
the process type was ALL and the base method was UNIQUAC, which
uses ideal gas and Henrys law, best tting the process conditions.
2.1. Reactor Separation Model (Conguration A). The kinetic
data for the transesterication reaction was incorporated from a previous
experimental study in our laboratory.11 The activation energy and the
pre-exponential factor were calculated using linear regression of the
experimental data reported. The kinetic parameters are shown in Table 2.
Table 2. Arrhenius Parameters for the Transesterication

Reaction11
pre-exponential factor (A) (s1)
activation energy (Ea) (J/mol)
0.5574
30700
Because results in ref 11 indicated higher biodiesel yields at higher

temperatures (150200 C), an isothermal operation with no pressure
drop at 160 C was chosen. Reaction temperatures higher than 150 C do
not signicantly aect the biodiesel yield but increase the cooling water
utility requirements. To obtain the optimized reactor temperature
operation, reactor simulations were run in the temperature range of
150200 C and, at 160 C, a maximum conversion of 99.81% was
achieved. For sizing the reactor, the Aspen reactor model conguration
provides length, diameter, and number of tubes as adjustable parameters.
This is a continuous process; therefore, while the reactor was sized varying
6850
Energy & Fuels
Article
Figure 6. Flow sheet units: reactor separation ow sheet (conguration A).

column. This 6:1 optimal ratio was chosen for best conversion after a set
of trial simulation runs of the CD column. A total holdup of 50 lb mol
was imposed for the liquid phase. The distillate purity was above 99%
mole fraction of methanol and could be recycled. The mixer functioned
to merge the recycle into the fresh feed. The total conversion for triolein
in the process is 99.63%, and the bottom stream from the RADFRAC
column was sent to a decanter to settle and separate glycerol from
methyl oleate.
The operation of the CD column at lower pressures results in very
little methanol in the bottom product (0.0044 mole fraction); this is a
notable benet of the CD operation. The bottom product from the CD
column is composed mostly of glycerol and methyl oleate, which are
easily separated by operating the decanter, owing to a signicant
dierence between their densities (glycerol density of 1.26 g/cm3 and
methyl oleate density of 0.87 g/cm3). The decanter temperature and
pressure conditions were set close to the bottom stream from the
RADFRAC column to minimize energy expenses while maintaining the
purity standards. At an operation of 150 C and 1 atm, its heat duty was
the lowest for a separation of 99% methyl oleate. A nal stream of
99.10% mole fraction purity for methyl oleate was obtained from the
decanter as well as another stream with a purity of 99.38% for glycerol.
These are very high product purity standards, and hence, additional
distillation columns were not required. The composition and ow rate of
all constituent streams for the reactor separation conguration are
shown in Table 5.
Table 3. Reactor Conguration

number of tubes
reactor total length (m)
single tube diameter (m)
reactor total volume (m3)
100
10.50
0.37
112.9
and 99.56% of FAME on the bottom stream (99.71% mass fraction). The
FAME distillation column operates at 0.3 atm (vacuum distillation) to
reduce the temperature of the product FAME stream (biodiesel starts
thermal degradation via isomerism, polymerization, and pyrolysis at
temperatures exceeding 275 C).18 Low-pressure distillation for biodiesel
has also been reported in the literature.8 Methanol separation from
biodiesel is utmost necessary to meet ASTM standards. Most biodiesel
standard allows only 0.2% (v/v) methanol in the nal product.19 Residual
methanol in the biodiesel fuel is a major environmental and health hazard
because of a number of reasons. Methanol is toxic (ingestion of 10 mL
causes permanent blindness), has cold-start problems and lower energy
density, and evaporates quickly when exposed to air. Excess methanol can
also make the fuel ammable and more dangerous to handle and store
besides, corroding metal components of the engine.13,19,20 The glycerol
purication column achieved a separation of 99.87% mole fraction for
methanol on the top stream (99.29% mass fraction), which is recycled
back into the mixer, and a purity of 98.77% of glycerol on the bottom
stream (99.53% mass fraction). The glycerol purication column operated
by 0.5 atm to have a product glycerol stream less than its boiling point,
287.71 C. This low-pressure distillation operation at 0.5 atm was also
featured in the work performed in ref 8. The composition and ow rate of
all constituent streams for the reactor separation conguration are shown
in Table 4.
2.2. CD Model (Conguration B). One of the most signicant
merits that a CD brings to a process is simplication of the ow sheet
and savings in equipment cost and operation. In this case, the biodiesel
process is intensied by removal of the plug-ow reactor and two
distillation columns from conguration A by a single CD column in
conguration B, where both reaction and separation occur. For the CD
conguration, the process equipment required was a mixer for enabling
the recycle of methanol into the reactor, a reactive distillation column
(RADFRAC) for the reaction and methanol separation, and a decanter
for glycerol and methyl oleate separation. The elements are shown in
Figure 7.
The CD column was modeled in Aspen via an equilibrium-based
rigorous two-phase fractionation model (RADFRAC) with a total
number of seven stages (ve trays, condenser, and reboiler) and a reux
ratio of 0.6. The column was operated at a pressure of 3 atm and a per
stage pressure drop of 0.1 atm. This is another advantage demonstrated
by a CD operation. Because the heat of the exothermic reaction is
consumed to separate out the products, the column is able to operate at
higher pressures while maintaining the reboiler product biodiesel at
temperatures less than its degradation temperature (275 C). Two
separate feed streams were added to the column. The lighter component
alcohol was fed at stage 6 close to the reboiler, and the heavier
component oil was fed at stage 2 close to the condenser, to enable the
reaction to take place between these stages. Design criteria were
specied for the ow rate of the fresh alcohol feed to maintain the 6:1
methanol/triolein ratio for the stream exiting the mixer and entering the
3. PROCESS COMPARISONS (COST AND ENERGY)

Both process congurations were optimized for desired biodiesel
ASTM purity standards while minimizing energy requirements.
The nal streams from both biodiesel process congurations A
and B yield the same mass ow rate (8900 lb mol h1) and
percentage purity (99.00 mol %) of biodiesel (methyl oleate).
Operating conditions for both columns were set to have less than
0.2% (v/v) methanol in the nal product to comply with ASTM
standards. The aim of this research is to compare the total capital
and operating costs between the two optimized process
congurations for obtaining the same purity and ow of biodiesel
to predict quantitatively the more cost-ecient process. To
achieve this, the Aspen process economic analyzer tool
(formerly, Aspen Icarus Process Evaluator) was used. The
process economic analyzer tool is most valuable to compare
competing technologies and/or evaluate alternative process
congurations that are able to calculate the preliminary size for
process equipment and generate operating and capital costs
using built-in design and cost models directly from simulation
data.
Tables 6 and 7 enlists the total capital costs, total operating
costs, equipment purchase costs, and yearly utility and raw
material costs for congurations A and B, respectively, for an
annual biodiesel production of around 10 million gallons/year.
These values were evaluated using generated mass and energy
results from Aspen process simulators, plugging in stream prices
and then mapping and sizing the equipment using the process
6851
6852
90.00
10.00
0.55
0.01
0.8949
0.0994
0.0055
0.0001
0.2445
0.7509
0.0043
0.0003
100.56
11792.31
491.68
30.09
10.00
0.7505
0.2495
0.0982
0.9018
40.09
9818.35
4.93
1.00
68.27
1.00
0.38
0.62
R-IN
25.00
1.00
Stream names are specied in Figure 2.
temperature (C)
pressure (atm)
vapor fraction
liquid fraction
component mole ow (lb mol h1)
methanol
triolein
glycerol
methyl oleate
component mole fraction
methanol
triolein
glycerol
methyl oleate
component mass fraction
methanol
triolein
glycerol
methyl oleate
total mole ow (lb mol h1)
total mass ow (lb h1)
volume ow (m3 h1)
FEED
0.1632
0.0014
0.0823
0.7532
100.56
11792.50
7.03
0.5972
0.0002
0.1048
0.2979
60.05
0.02
10.54
29.96
1.00
160.00
15.00
R-OUT
Table 4. Flow Sheet for the Reactor Separation: Conguration Aa
0.0148
0.0016
0.0923
0.8913
44.54
9961.42
5.62
0.1031
0.0004
0.2241
0.6724
4.59
0.02
9.98
29.95
1.00
125.00
1.00
FL-BOT
0.0564
4.09 1011
0.9429
0.0007
11.59
965.88
0.38
0.1466
3.85 1012
0.8532
0.0002
1.70
4.47 1011
9.89
2.32 103
1.00
125.00
1.00
DEC-BOT
0.0103
0.0018
0.0010
0.9869
32.95
8995.55
5.10
0.0877
0.0006
0.0028
0.9089
2.89
0.02
0.09
29.94
1.00
125.00
1.00
DEC-TOP
0.0001
0.0018
0.0010
0.9971
30.08
8903.59
5.81
0.0007
0.0006
0.0031
0.9956
0.02
0.02
0.09
29.94
1.00
258.17
0.30
FAME
0.0043
1.81 1021
0.9953
0.0005
10.01
914.96
0.37
0.0121
1.87 1022
0.9877
0.0001
0.12
1.87 1021
9.89
1.49 103
1.00
188.44
0.50
GLYCEROL
0.9706
1.65 105
0.0278
0.0016
56.03
1831.08
830.28
0.9900
6.08 107
0.0099
0.0002
55.47
3.41 105
0.55
0.01
125.00
1.00
1.00
RECYCLE1
1.0000
7.11 106
4.11 107
8.58 108
2.87
91.96
0.05
1.0000
2.57 107
1.43 107
9.27 109
2.87
7.38 107
4.11 107
2.66 108
1.00
36.80
0.30
RECYCLE2
0.9929
7.76 1010
0.0022
0.0048
1.58
50.92
0.03
0.9987
2.83 1011
0.0008
0.0005
1.58
4.47 1011
1.24 103
8.31 104
1.00
47.94
0.50
RECYCLE3
Energy & Fuels

Article
Energy & Fuels
Article
product stream price for technical-grade glycerol was set as

0.9 dollars/lb22 (the glycerol product stream in both congurations is more than 98.7 mol % in purity). The total capital
corresponds to the investment required for purchase of equipment, costs of labor and materials (direct installation costs), costs
for site preparation and buildings, and certain other costs (indirect
installation costs). It also includes costs for land, working capital,
and o-site facilities. The total operating costs include labor,
maintenance, utilities, and raw material costs. Detailed information
regarding parameters and evaluation basis for operating and capital
costs can be found via generating the economics report le from
the Aspen process economic analyzer toolbar.
The implementation of heterogeneous catalyst systems for the
biodiesel simulations in this study deserves special attention.
Despite slower reaction kinetics and higher costs, they constitute
an interesting area of study in the biodiesel process. Use of
heterogeneous catalyst systems would signicantly simplify the
separation process and reduce equipment and utility costs.
Figure 7. Flow sheet units: CD ow sheet (conguration B).
economic analyzer tool. The feed stream prices were set as

26 cents/lb for soybean oil and 24 cents/lb for methanol.21 The
Table 5. Flow Sheet for the CD: Conguration Ba
stream
OIL
temperature (C)
pressure (atm)
vapor fraction
liquid fraction
component mole ow (lb mol h1)
methanol
triolein
glycerol
methyl oleate
component mole fraction
methanol
triolein
glycerol
methyl oleate
component mass fraction
methanol
triolein
glycerol
methyl oleate
total mole ow (lb mol h1)
total mass ow (lb h1)
volume ow (m3 h1)
a
ALCOHOL
ALCOHOL2
DISBOT
DISTOP
FAME
GLYCEROL
25.00
3.20
25.00
1.00
62.00
3.70
273.37
3.70
95.43
3.00
150.00
1.00
150.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
30.07
60.00
2.88 103
0.06
1.33 103
0.18
0.04
9.96
29.89
29.93
2.88 103
0.06
1.33 103
0.12
0.04
0.12
29.89
0.06
4.57 1011
9.85
1.70 103
1.00
0.9989
4.79 105
0.0010
2.21 105
0.0044
0.0009
0.2487
0.7460
0.9978
0.0001
0.0021
4.42 105
0.0039
0.0012
0.0039
0.9910
0.0060
4.61 1012
0.9938
0.0002
1.00
0.9955
0.0013
0.0030
0.0002
60.07
1931.26
1.17
0.0006
0.0033
0.0935
0.9027
40.07
9817.91
6.51
0.9910
0.0026
0.0060
0.0004
30.00
967.85
0.62
0.0004
0.0036
0.0012
0.9947
30.16
8908.80
5.17
0.0021
4.45 1011
0.9973
0.0006
9.91
909.10
0.35
10.00
1.00
1.00
10.00
8854.30
2.83
30.07
963.42
0.55
Stream names are specied in Figure 3.
Table 6. Detailed Cost Analysis for Optimum Design and Operating Conditions for the Reactor Separation Conguration
(Conguration A)
conguration A: reactor + distillation
optimal
specication
optimal operation condition
heat duty (kW)
volume
(112.9 m3)
volume
(3.27 m3)
volume
(1.80 m3)
temperature (160 C),

pressure (15 atm)
pressure (1 atm)
pressure (1 atm)
127.63544
FAME distillation
column
stages (5),
feed (3)
mole RR (0.5), pressure

(0.3 atm)
glycerol distillation
column
stages (3),
feed (2)
mole RR (0.5), pressure

(0.5 atm)
equipment
reactor
ash separator
decanter
62.4769071
14.899454
20.044019
386.883111
11.698744
28.4992635
utility cost ($/year)

raw material cost ($/year)
equipment utility
cost ($/year)
total operating
cost ($/year)
equipment
purchase cost ($)
9618.60
204800
4708.26
19200
1122.82
16100
30667.00
3029.32
25509398.00
70500
total capital
cost ($)
5871380.00
52800
49145.00
22222700.00
6853
Energy & Fuels
Article
Table 7. Detailed Cost Analysis for Optimum Design and Operating Conditions for the CD Conguration (Conguration B)
conguration B: CD
optimal operation condition
heat duty
(kW)
equipment utility
cost ($/year)
equipment
optimal specication
decanter
volume (1.80 m3)
temperature (150 C), pressure

(1 atm)
415.90
13423.32
RADFRAC
stages (7), feed (2, 6)

(8.53 m high)
mole RR (0.6), pressure (3 atm), per

stage pressure drop (0.1 atm)
210.30
620.56
26815.98
utility cost ($/year)

raw material cost ($/year)
total operating
cost ($/year)
equipment
purchase cost
($)
total capital
cost ($)
13500.00
25001100.30
89700.00
3439170.00
40239.30
22207200.00
Table 8. Utility Usage and Costs Corresponding to Individual Process Equipment for the Reactor Separation Conguration
(Conguration A)
energy consumption/h
(kW)
mixer
reactor
ash separator
decanter
glycerol column
FAME column
total heating
total cooling
net energy
requirement
total utilities cost
($/year)
total energy fraction
price fraction ($/year)
127.63544
62.4769071
14.899454
11.698744
28.4992635
20.044019
386.883111
477.8592816
174.277657
652.1369386
heating
fraction
cooling
fraction
equipment heating cost

($/year)
0.73
0.13
equipment cooling cost

($/year)
equipment total utility cost

($/year)
9618.60
1122.82
881.62
9618.60
4708.26
1122.82
3029.32
1510.52
30667.00
4708.26
0.09
0.07
0.06
2147.70
0.12
0.81
29155.49
49145.00
heating
0.73
36011.45
cooling
0.27
13133.55
Figure 8. Liquid-phase concentrations of each component in the CD column (conguration B). The transesterication reaction (demonstrated by a
decreasing concentration of triolein) mostly takes place on trays 13.
considerably benet the cost of the overall process. Other merits

include higher selectivity, water tolerance, and lifetime of the
catalyst. In this work, we try to approximate the catalyst
Because glycerol is a valuable byproduct of the biodiesel

production process, we believe that a relatively purer supply of
glycerol from use of heterogeneous catalyst systems would
6854
Energy & Fuels
Article
preparing the catalyst bed. Hence, the total catalyst costs were
approximately 190 751 dollars for conguration A and 301 210
dollars for conguration B. These values are rough estimates for
catalyst requirements and costs and will vary depending upon
packing, bed characteristics, equipment geometry, temperature and
pressure conditions, and ow rates of the process. The objective of
calculating catalyst requirement in this research is to gauge an idea of
the probable costs associated with changing the biodiesel
production process to heterogeneous catalysis because no cost
estimate for a heterogeneous catalyzed process for biodiesel was
available in the literature. It is of note that the catalyst requirement
for conguration B appears to be more than that for conguration A.
Further investigation of composition and temperature proles for
the CD column (Figures 8 and 9) demonstrate that most of the
reaction is taking place between trays 1 and 3. Hence, in actual
running of the CD column, the required catalyst loading for the
bottom trays should be less than the calculated value and both
congurations would have relatively closer catalyst requirements.
Results demonstrate that the CD conguration (B) is signicantly
more economical compared to the reactor separation conguration
(A) in terms of capital and utility costs. There is only meagerly
savings in terms of total operating and raw material costs per year.
The total capital cost in dollars for the CD conguration (B) is
3.44 million dollars (41.42% less) compared to 5.87 million dollars
for the reactor separation conguration (A). These capital costs are
in agreement with reported capital costs for 10 million gallon annual
production biodiesel plants that use soybean oil as feedstock using
homogeneous catalysts.6,23,24 The total operating cost in dollars per
year for the CD conguration (B) is 24.95 million dollars (1.46%
less) compared to 25.32 million dollars for the reactor separation
conguration (A). Numbers for operating costs for an annual 10
million gallon biodiesel soybean oil facility closely match reported
literature.6,25 Because we are working at the same ow rates and
achieving the same conversions, the raw material costs per year for
both congurations are nearly identical. The utility costs per year for
the CD conguration (B) are 18.12% lower compared to the reactor
separation conguration (A). Aspen used built-in heat integration
techniques (pinch technology) to minimize the utility costs that can
be accessed using the energy analysis icon on the analysis toolbar.
Utility usage and utility costs corresponding to each equipment in
ow sheet congurations A and B are listed in Tables 8 and 9,
respectively.
The production cost per gallon of biodiesel is a signicant
factor to predict the protability of the production process. From
the annual production capacity and the total operating cost per
year for the plant, the production cost per gallon of biodiesel was
calculated. The calculations are shown in Table 10. The
production cost per gallon of biodiesel for both congurations
Figure 9. Temperature prole (liquid phase) in the CD column

(conguration B). A steeper temperature prole between trays 1 and 3
indicates the region of the exothermic transesterication reaction.
requirements and associated costs into the total annual operating

costs of the whole process. It must be noted that our simulations are
modeled on kinetic parameters taken from the performance of 20%
calcium oxide supported on Al2O3 (20% calcium oxide by weight,
with 80% alumina) as the solid base (discussed in section 211).
Assuming a catalyst life of 1 year and catalyst loading of 3 wt %
catalyst/oil ratio (maximum ester yield at this ratio11), costs
associated with this heterogeneous catalyst system are added to the
economic analysis results generated from the Aspen process
economic analyzer tool. For conguration A, the residence times
associated with the packed-bed reactor were used to approximate
the triolein amount at any specic time inside the reactor and,
accordingly, the catalyst requirement was calculated. For conguration B, the liquid holdup on each stage, feed composition, and
ow rate were used to approximate the triolein amount and the
corresponding catalyst requirements. Price specications for
alumina Al2O3 (Brockmann I, activated, 150 mesh size) and
calcium nitrate terahydrate Ca(NO3)24H2O for synthesizing
calcium oxide were taken from Sigma-Aldrich. The associated
catalyst requirement for conguration A (the reaction + distillation
process) is around 1907.51 kg/year, whereas for conguration B
(CD process), it is 3012.0975 kg/year. As of August 2013, the
Sigma-Aldrich price for alumina Al2O3 (Brockmann I, activated, 150
mesh size) is 387 dollars for 5 kg and the Sigma-Aldrich price for
calcium nitrate terahydrate Ca(NO3)24H2O is 268 dollars for
2.5 kg. Using these prices, the cost of a catalyst system comprising
20% calcium oxide and 80% alumina by weight comes out to be
72.64 dollars/kg. Because batch and bulk costs for chemicals vary
signicantly, the catalyst was assumed to have an average cost of 100
dollars/kg. This price also allows for some compensation for the loss
in material while synthesizing the catalyst and costs involved in
Table 9. Utility Usage and Costs Corresponding to Individual Process Equipment for the CD Conguration (Conguration B)
energy consumption/h
(kW)
decanter
distillation column
total heating
total cooling
net energy requirement
total utilities cost ($/year)
total energy fraction
price fraction ($/year)
415.90298
210.29701
620.559298
620.559298
626.19999
1246.759288
40239.30
heating
0.50
20028.62
heating
fraction
cooling
fraction
equipment heating cost

($/year)
0.66
0.34
1.00
equipment cooling cost

($/year)
equipment total utility cost

($/year)
13423.32
6787.36
13423.32
26815.98
20028.62
cooling
0.50
20210.68
6855
Energy & Fuels
Article
gratefully acknowledged. Mateus Lenz Leite acknowledges

nancial support for an exchange scholarship provided by the
Brazilian National Council for Scientic and Technological
Development (CNPq) through Science without Borders
Program (CsF). Aashish Gaurav acknowledges a Trillium
scholarship for Graduate Studies provided by the Ontario
Government.
Table 10. Per Gallon Production Cost of Biodiesel for

Congurations A and B
reactor + distillation
(conguration A)
CD
(conguration B)
5871380.00
22222700.00
3439170.00
22207200.00
49145.00
190751
25509398.00
7214920
40,239.30
301210
25001100.30
7172290
8903.59
0.99
4027.089
870
40548.62
2.39
8908.80
0.99
4019.67
870
40473.92
2.36
1.71
1.67
total capital cost (USD)

total raw materials cost
(USD/year)
(methanol + soybean oil)
total utilities cost (USD/year)
total catalyst cost
total operating cost (USD/year)
total product sales (USD/year)
(glycerol)
total FAME stream ow (lb/h)
FAME mass fraction
total FAME mass (kg/h)
density (kg/m3) at 25 C and 1 atm
total volume (m3/year)
price without considering glycerol
revenue ($/gallon)
price considering glycerol revenue
($/gallon)
(1) Kreutzer, U. Manufacture of fatty alcohols based on natural fats and

oils. J. Am. Oil Chem. Soc. 1984, 61 (2), 343348.
(2) Hill, J.; Nelson, E.; Tilman, D.; Polasky, S.; Tiffany, D.
Environmental, economic, and energetic costs and benefits of biodiesel
and ethanol biofuels. Proc. Natl. Acad. Sci. U. S. A. 2006, 103 (30),
1120611210.
(3) Ng, F. T. T.; Rempel, G. L. Catalytic distillation. Encyclopedia of
Catalysis; John Wiley and Sons, Inc.: Hoboken, NJ, 2002.
(4) Yokoyama, S. The Asian Biomass Handbook; Japan Institute of
Energy: Tokyo, Japan, 2008.
(5) Ma, F.; Hanna, M. A. Biodiesel production: A review. Bioresour.
Technol. 1999, 70 (1), 115.
(6) Haas, M. J.; McAloon, A. J.; Yee, W. C.; Foglia, T. A. A process
model to estimate biodiesel production costs. Bioresour. Technol. 2006,
97 (4), 671678.
(7) Myint, L.; El-Halwagi, M. Process analysis and optimization of
biodiesel production from soybean oil. Clean Technol. Environ. Policy
2009, 11 (3), 263276.
(8) Zhang, Y.; Dube, M. A.; McLean, D. D.; Kates, M. Biodiesel
production from waste cooking oil: 1. Process design and technological
assessment. Bioresour. Technol. 2003, 89 (1), 116.
(9) Zhang, Y.; Dube, M. A.; McLean, D. D.; Kates, M. Biodiesel
production from waste cooking oil: 2. Economic assessment and
sensitivity analysis. Bioresour. Technol. 2003, 90 (3), 229240.
(10) You, Y.-D.; Shie, J.-L.; Chang, S.-Y.; Huang, S.-H.; Pai, C.-Y.; Yu,
Y.-H. Economic cost analysis of biodiesel production: Case in soybean
oil. Energy Fuels 2007, 22 (1), 182189.
(11) Pasupulety, N.; Gunda, K.; Liu, Y.; Rempel, G. L.; Ng, F. T. T.
Production of biodiesel from soybean oil on CaO/Al2O3 solid base
catalysts. Appl. Catal., A 2013, 452, 189202.
(12) Oliveira, C. F.; Dezaneti, L. M.; Garcia, F. A. C.; de Maced, J. L.;
Dias, J. A.; Dias, S. C. L.; Alvim, K. S. P. Esterification of oleic acid with
ethanol by 12-tungstophosphoric acid supported on zirconia. Appl.
Catal., A 2010, 372 (2), 153161.
(13) Van Gerpen, J.; Shanks, B.; Pruszko, R.; Clements, D.; Knothe, G.
Biodiesel Production Technology; National Renewable Energy Laboratory
(NREL): Golden, CO, 2004; NREL/SR-510-36244.
(14) Fogler, H. S. Elements of Chemical Reaction Engineering, 4th ed.;
Prentice Hall: Upper Saddle River, NJ, 2006.
(15) Baughman, W. F.; Jamieson, G. S. The chemical composition of
soya bean oil. J. Am. Chem. Soc. 1922, 44 (12), 29472952.
(16) Smith, W. B. The composition of soy-bean oil. J. Ind. Eng. Chem.
1922, 14 (6), 530530.
(17) Krawczyk, T. Biodiesel. Int. News Fats, Oils Relat. Mater. 1996, 7
(8), 800.
(18) Lin, R.; Zhu, Y.; Tavlarides, L. L. Mechanism and kinetics of
thermal decomposition of biodiesel fuel. Fuel 2013, 106, 593604.
(19) Bondioli, P. The Biodiesel Handbook. By Gerhard Knothe, Jon
Van Gerpen and Jurgen Krahl (Eds.). Biotechnol. J. 2007, 2 (12), 1571
1572.
(20) Berrios, M.; Skelton, R. L. Comparison of purification methods
for biodiesel. Chem. Eng. J. 2008, 144 (3), 459465.
(21) Methanex Corporation. Methanol PriceMethanex Regional
Posted Contract Prices; http://www.methanex.com/products/
methanolprice.html (accessed July 7, 2013).
(22) ICIS. ICIS Indicative Chemical Prices AZ; http://www.icis.com/
chemicals/channel-info-chemicals-a-z/ (accessed June 17, 2013).
(23) Van Gerpen, J. H. Biodiesel economics. Proceedings of the Oilseeds
and Biodiesel Workshop; Billings, MT, Jan 910, 2008.
comes out to be around 2.32.4 dollars/gallon. After accounting

for the glycerol sales from the product stream, the production
cost comes out to be around 1.61.7 dollars/gallon for both of
the congurations. Values for production costs are in agreement
with ref 6 and 26 and several white papers published. It may be
noted that, in the literature,6 published in 2005, a biodiesel
production cost of around 2 dollars/gallon was predicted. Our
predicted costs are very close, if the cumulative ination factor
between years 2005 and 2013 (around 15.06% as predicted by
Statistics Canada) is considered between the raw material costs.
4. CONCLUSION
The present research established the commercial feasibility of replacing
the traditional reactor separation technology by the CD technology.
The Aspen Plus process models developed can calculate optimal
process conditions and process economics for biodiesel production. A
comparison of conventional reactor separation and CD ow sheet
simulation data predicts that CD can lead to signicant savings in
capital and utility costs. One of the hallmarks of this research is the
investigative study of heterogeneous catalyst systems for the biodiesel
process and an approximation of the associated catalyst requirements
and corresponding costs. Another advantage of CD conguration was
the possibility of high-pressure operation while maintaining low
product stream temperatures. The production cost per gallon of
biodiesel calculated is another suitable tool to predict the protability of
the operation. The process model developed in Aspen Plus is exible
to accommodate higher ow rates for scale-up of operations, add or
remove stages of operation, and predict associated capital and
production costs. One can also manipulate the feedstock and methanol
price in the model to obtain dierent protability and cost scenarios.
REFERENCES
AUTHOR INFORMATION
Corresponding Author
*E-mail: fttng@uwaterloo.ca.
Notes
The authors declare no competing nancial interest.
ACKNOWLEDGMENTS
Financial support from the Natural Sciences and Engineering
Research Council of Canada (NSERC) for this research is
6856
Energy & Fuels
Article
(24) Tyson, K. S., Biodiesel technology, economics and case studies.

Proceedings of the NAEMI Biomass and Business Training Workshop;
Spokane, WA, May 1519, 2006.
(25) Fortenbery, T. R. Biodiesel Feasibility Study: An Evaluation of
Biodiesel Feasibility in Wisconsin; University of WisconsinMadison:
Madison, WI, 2004.
(26) Wisner, R. Biodiesel economicsCosts, tax credits and coproduct. AgMRC Renewable Energy Newsletter, June 2009; http://www.
agmrc.org/renewable_energy/biodiesel/biodiesel-economics-coststax-credits-and-co-product (accessed July 7, 2013).
6857
Article
pubs.acs.org/IECR
Production of Isooctane from Isobutene: Energy Integration and

Carbon Dioxide Abatement via Catalytic Distillation
Behnam M. Goortani, Aashish Gaurav, Alisha Deshpande, Flora T. T. Ng,* and Garry L. Rempel
Department of Chemical Engineering, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada
ABSTRACT: Isooctane is a valuable octane enhancer for gasoline and the primary component of aviation gasoline, also known
as Avgas, because of its high antiknock quality. Conventional industrial processes for isooctane production involve the steps of
dimerization of isobutene, dimer separation, and hydrogenation. The ecacy of catalytic distillation (CD) and its merits, in terms
of energy savings and reduction of greenhouse gas emissions, for the production of isooctane are quantitatively presented. The
feed considered for the isooctane production is composed of isobutene (C4) and inerts (isopentane) produced in reneries as
byproducts of steam cracking of naphtha and light gas oil. Process ow sheets for the two routes for the production of isooctane,
with and without CD, are modeled. The conventional industrial ow sheet composed of a dimerization reactor, distillation
column, and a hydrogenation reactor (conguration A), is simulated using Aspen Plus. The intensied process ow sheet
comprising a CD column for the dimerization, hydrogenation, and separation (conguration B) is modeled using gPROMS. A
validated, nonequilibrium, three-phase model is developed in a gPROMS environment and is used to quantify the energy savings
and reduction of carbon dioxide emissions achieved using a CD column for the intensied process. Results demonstrate CD to
be a promising candidate to replicate the conversions and product purity obtained in the conventional process while resulting in
signicant energy savings, more ecient utilization of isobutene feed, and reduced carbon dioxide emissions.
1. INTRODUCTION
Recent renery technologies have been directed at producing
high-octane gasoline-blending components that are essential to
increase the compliance of motor gasolines with the quality
specications and projected quantity demand.1,2 The phase out
of methyl-tert-butyl ethanol (MTBE) in North America,
increased demand for middle distillates (kerosene and diesel)
in comparison to gasoline fractions, implementation of the
latest European fuel specications, and adoption of cleaner and
more stringent fuel quality specications worldwide have
necessitated eorts toward the production of greater quantities
of high-octane blending components for gasoline that do not
contain aromatics, benzene, olens, and sulfur.24
In regard to these recent oil rening developments, isooctane
has emerged as one of the leading gasoline additives on account
of its high octane number, low vapor pressure, and zero content
of aromatics and sulfur.57 Isooctane is also the primary
component of aviation gasoline (Avgas) because of its knock
resistance and high compression ratio.8,9 Isooctane production
technology oers the scope of utilizing surplus isobutene in
reneries due to the MTBE phase-out and increased renery
revenues as conversion of light olens into gasoline blends
results in increased gasoline sales. Isooctane use and production
are expected to rise signicantly in the near future.
The conventional process for isooctane manufacturing
involves dimerization of isobutene in a xed bed reactor with
a supported acid catalyst, followed by hydrogenation in a
continuous stirred tank reactor with hydrogenation catalysts.7,10,11 There are various isooctane processes available
commercially, varying in terms of reaction conditions, reactant
conversion, and catalyst type, namely, CDIsoether, InAlk,
Selectopol, SP-Isoether, NExOCTANE, etc.7,12,13 Most conventional processes for isooctane production operate at low
conversions (2060%, with conversions diering according to
2015 American Chemical Society
the catalyst and reactor design) to prevent catalyst sintering due

to the highly exothermic reaction and to avoid the formation of
higher oligomers that may result from the increased
concentration of dimer in the reactor.1416 Conversions higher
than 60% per pass are often not feasible and rare in industry
because of these constraints. Our objective is to quantify the
merits of catalytic distillation (CD) over the conventional
reactor followed by separation process. We model the
conventional process at dierent conversions to get an accurate
comparison of results between a conventional and CD process.
The detailed process owsheet diagram for the conventional
process is shown in Figure 1.
Process intensication is a principle of green engineering
whereby two or more unit operations are combined into a
single unit operation with the benecial result of increased
eciency, reduced operating and capital costs, and a reduction
of waste and recycle streams. The sensitivity of the isooctane
process performance to various parameters such as type of
catalyst, process conditions, and reactor conguration opens
intriguing options for process design and intensication. CD is
a green engineering technology that combines the functionalities of a chemical reactor into a distillation column by
immobilizing solid catalyst particles within discrete reactive
sections.17 This distinctive feature of CD to simultaneously
carry out the chemical reaction and the product separation and
purication within a single-stage operation results in signicant
capital savings due to process intensication. The continuous
removal of product from the reactive section via the distillation
Received:
Revised:
Accepted:
Published:
3570
August 12, 2014

March 18, 2015
March 27, 2015
March 27, 2015
DOI: 10.1021/ie5032056
Ind. Eng. Chem. Res. 2015, 54, 35703581
Industrial & Engineering Chemistry Research
Article
Figure 1. Simplied process ow sheet diagram for the conventional isooctane process.
situ separation of products in CD utilizing the reaction heat is

the reduction of the formation of higher oligomers of isobutene
such as dodecene (side reactions), thus resulting in higher
product selectivity. CD is, therefore, a very promising option
for isooctane production. The objective of this research is to
quantity some of the above merits.
The eectiveness of a CD process compared to conventional
non-CD processes has been studied for various reactions by
many researchers.1820 The intensied process using CD
results in higher conversion as well as reduced total condenser
and reboiler duties. However, to the knowledge of the authors,
in the literature to date, no systematic, quantied work has
been done toward assessing the savings of energy and reduction
of greenhouse gases when a reactor and distillation column are
replaced with a CD column for the process of isooctane
production from isobutene. This is an industrial reaction of
great signicance to downstream renery operations because
butene isomers are obtained as byproducts of the steam
cracking of naptha and light gas oil. To achieve a detailed
comparison and quantication of the merits of CD, we have
modeled both the conventional isooctane production process
(reactor and distillation) and the intensied CD process. The
conventional isooctane process is modeled in Aspen Plus, one
of the most extensively used chemical engineering packages.
The intensied CD process is modeled using gPROMS, an
equation-oriented advanced modeling and simulation tool.
Though results from process simulations and actual process
operations dier to some extent, simulation software tools
generally provide quite reliable and accurate information on
process operations and the inuence of process parameters
because of their comprehensive thermodynamic packages, rich
component data, and astute calculation techniques. Hence, the
modeling results could provide comparison between performance of competing technologies in real operations.
action can also lead to higher conversion and selectivity,

particularly for reactions that are equilibrium-limited because
the products formed are immediately removed by distillation,
thereby preventing the products from participating in other
undesired side reactions. Other potential advantages of CD
include the mitigation of catalyst hot spots, improved
temperature control, and enhanced energy integration due to
conduction of an exothermic chemical reaction in a boiling
medium with in situ separation via distillation.
CD nds applications primarily for reactions in which the
dierence in volatilities of the reactants and the products is
suciently high so that they may be feasibly separated by
distillation. In addition, the reaction should be exothermic and
should take place in the liquid phase. The isooctane production
process is an ideal candidate for CD operation and ts the
design criteria perfectly. The dimer products dier substantially
in volatility, which makes separation by distillation favorable
(see physical properties listed in Table 1). Moreover, the
exothermicity of both the dimerization and hydrogenation
reactions involved in the synthesis of isooctane from isobutene
reported in the literature5,11 greatly favors a CD operation as it
would fulll most of its energy requirements for product
separation from the exothermicity of the reactions. This would
lower the heat duty of the reboiler. An added advantage of in
Table 1. Boiling Point and Density of Isobutene and Its
Dimer Product Isooctane
isobutene
molecular weight
(g/mol)
boiling point (C)
density (g/cm3)
isooctene
(dimer)
56.107
114.23
7.2 C (1.013 bar)

0.626 (at boiling point,
1.013 bar)
99 C
0.69
3571
DOI: 10.1021/ie5032056
Ind. Eng. Chem. Res. 2015, 54, 35703581
Article
octane number fuel, isooctane. The dimerization reaction is

highly exothermic with a heat of the reaction of H = 82.9
kJ/mol.5,16 A RSTOIC reactor model in Aspen Plus also gave
similar results. The RSTOIC model predicted the heat of
reaction to be 76.2 kJ/mol at 1 atm and 300 K for the
isobutene dimerization reaction. Hydrogenation reactions of
alkenes are known to be exothermic, but this was also veried
(estimated around 109.55 kJ/mol) by running a RSTOIC
Aspen model. The exothermicity of both reactions makes CD
an ideal candidate for in situ heat utilization toward separation
of products.
The reaction kinetic data for the dimerization and the
hydrogenation reactions in the isooctane production process
were incorporated from previous experimental studies carried
out in our laboratory25 using a nickel sulfate and a Pd catalyst
separately supported on -Al2O3.24 These kinetic data
correspond to the liquid-phase oligomerization and hydrogenation of isobutene obtained in a 300 mL Parr autoclave
reactor. The dependence of the reaction rate constant on the
reaction temperature was based on the Arrhenius equation, k =
Ae(E/RT) where the reaction constants k1 and k2 for the
dimerization and hydrogenation, respectively, are given below.
Dimerization Reation:
The process congurations for the conventional process and

the catalytic distillation process have been modeled in dierent
software systems (Aspen Plus and gPROMS) on account of
modeling constrainsts. The CD pilot plant runs were modeled
using a validated, three-phase, rate-based nonequillibrim model
coded in gPROMS. The current gPROMS library does not
have reactor models that can handle components in both the
liquid and vapor phase. Current reactor models in gPROMS
can handle only gas-phase reactions. The gPROMS library is
also presently devoid of distillation column models that
quantify reactor and condenser duties for a given product
purity. The authors also note that the model equations for the
reactor and distillation models remain the same in dierent
software libraries. The dierences in software results arise from
thermodynamic packages for calculation of physical properties,
component data, and calculation techniques (solvers). These
dierences would not inuence the results appreciably.
There are previous literature reports on quantication of
energy savings obtained in a CD process against the
conventional processes for other reaction systems.18,2123
Reference 18 reports a reduction of 2040% in utility
requirements for the synthesis of methyl tert-butyl-ether when
the conventional process congurations were replaced by CD
congurations. Energy requirements were also stated to get
reduced by 25% on switching to CD technologies in real
MTBE production units in ref 22. For the synthesis of dimethyl
ether, CD has been reported to bring about 20% reduction in
annual utility costs.21 While refs 18, 21, and 23 report
comparisons based on Aspen Plus simulations, ref 22 reports
energy savings brought about by CD in real renery unit
operations. It is anticipated that energy savings should be much
higher when highly exothermic reaction systems such as
oligomerization and hydrogenation reactions are carried out
in a CD conguration. This is indeed the result we obtained in
this study.
In this study, results from both process congurations
(conventional process versus CD) are compared at the same
product purity (0.36 mole fraction of isooctane) on the basis of
per kilogram of product.24,25 A comparison of the utility
requirements and isobutene feedstock utilization is also
performed. The savings in energy requirements and the
elimination of isobutene waste are then quantied to relate
the eectiveness of CD as compared to the conventional
process.
This study uses a validated, exible, three-phase, nonequilibrium model coded in gPROMS using overall masstransfer coecients for depicting the CD column performance
(discussed in section 3.2). This research is hence able to
compare the energy consumption and proles of temperature
and composition of the isooctane process with and without the
use of a CD column.
rCC H =
4 8
22184
k1 = 175.1818 exp
RT
k1
k2
dimerization
C8H16 + H 2 C8H18
hydrogenation
(3)
(4)
The units of C and rate of reaction, r, are mol/s and mol/s/kgcat, respectively.
Hydrogenation Reaction:
rCC H =
8 18
d
[CC8H18] = k 2xC8H18 0.3PH2 0.33
dt
20678
k 2 = 92.4788 exp
RT
(5)
(6)
The units of PH2 and rate of reaction, r, are psig and mol/s/gcat, respectively.
3. PROCESS SIMULATION
3.1. Conventional Reactor Separation Process (Conguration A). Figure 1 shows a simplied process ow
diagram for the isooctane production process in a conventional
non-CD process. The feed, composed of isobutene and
isopentane as an inert, is rst dimerized to isooctene in the
reactor. The more volatile nonreacted monomer (isobutene) is
then separated from the heavier dimer product (isooctene) in
the distillation column and sent back to the reactor. The
isooctene then enters the hydrogenation reactor, where the
hydrogenation reaction occurs and the nal hydrogenated
product, isooctane, is obtained.
The owsheet diagram in Figure 1 is simulated in Aspen
Plus. The main processing equipment units are a mixer, a plugow xed bed reactor (PFR), a distillation column, and a stirred
tank reactor (CSTR). The mixer functions to enable the recycle
of the overhead unreacted monomer back into the reactor. The
column functions to separate the lower molecular weight
volatiles C-4 and C-5 hydrocarbons from the isooctene. The
stirred tank reactor serves as the hydrogenation reactor for
conversion of isooctene to isooctane.
2. REACTION: ISOBUTENE TO ISOOCTANE

The reaction system selected for isooctane production is the
dimerization of isobutene to isooctene with subsequent
hydrogenation to isooctane (illustrated in eqs 1 and 2).
2C4 H8 C8H16
d
[CC4H8] = 2k1CCC H 3.186
4 8
dt
(1)
(2)
These reactions are of considerable industrial importance

because a lower quality fuel, isobutene, is converted to a high
3572
DOI: 10.1021/ie5032056
Ind. Eng. Chem. Res. 2015, 54, 35703581
Article
Figure 2. Simulation of the conventional reaction separation technology (conguration A) in Aspen Plus environment.
In an actual renery operation, the recycle ratio is often

varied depending on the product requirements. The actual
recycle ratio for the NExOCTANE process was not available in
the literature. Therefore, the simulation for conguration A was
run under varying recycle ratios, and the results were recorded
for comparison with the CD model (conguration B). The CD
model was run at total recycle because our experimental CD
data was obtained at total reux. For the conventional process,
we have simulated (conguration A) at dierent conversions
and dierent recycles. All cases for conguration A were set so
as to produce isooctane at around 0.36 molar fraction of the
hydrogenated product (iso-octane) in the nal product stream
because our CD experiments produced isooctane at this purity.
The CD model described in the next section also produces a
0.36 molar fraction of iso-octane in the reboiler; therefore, a
comparison of the energy requirements per kilogram of product
is justied. As expected, results for the conventional reaction
separation process showed signicantly larger cold utility
requirements due to the intense cooling water requirements
for cooling the dimerization and hydrogenation reactors (Table
2).
3.2. Catalytic Distillation Process (Conguration B). A
23 ft (7 m) pilot CD process unit (Figure 3) was used for
obtaining the CD process data.19,20 The CD column is
composed of two reaction zones and three separation zones.
Isobutene as a feed is injected just below the dimerization zone.
The dimerization reaction is exothermic, and the in situ
reaction heat is utilized to enhance separation of volatiles,
aiding heat mitigation and temperature control of the column,
thus cutting down cooling water requirements and preventing
catalyst sintering. The separation of the volatiles and the
products also minimizes consecutive oligomerization reactions.
The less volatile and denser dimerized product, isooctene,
moves downward to the hydrogenation zone, where it reacts
with hydrogen to produce isooctane. The hydrogenation
reaction is highly exothermic; therefore, the produced heat in
the hydrogenation section further aids vaporization of volatile
components and reduces the reboiler load. The nonreacted
isobutene is stripped in the lower separation zone of the
column, and a pure isooctane product leaves the column. The
CD column in our laboratory functions at total reux; the
nonreacted isobutene, isopentane, and the trace amounts of
evaporated isoctane and isooctene are essentially totally
condensed in the condenser and returned to the column.
Figure 2 shows the detailed Aspen Plus simulation

environment for conguration A. To model the conventional
reaction separation process, the process ow diagram, streamow rates, and equipment specications were based on data
available in the literature for the NExOCTANE process for the
production of isooctane.13 The Aspen process model uses the
kinetic model for the catalyst system discussed in section 2. The
idea presented here is to compare the conventional process
owsheet and the CD process owsheet for the isooctane
process for the same catalyst system.
A constant feed with a molar rate of 1.5 kg/sec
(corresponding to a feed rate of approximately 11 880 lb/h
of isobutene feed similar to the feedrate of around 13 700 lb/h
as stated for the NExOCTANE process13) is fed to the reactor
in an equimolar ratio of isobutene and isopentane. Isopentane
represents the inert feed component which in a real renery
comprises C-3, C-4, and C-5 alkanes. A jacketed tubular (plugow) reactor is used in the model. The reactor functions at 25
atm and 370410 K (temperature varies along the length of the
reactor) and produces isobutene at varying conversions (25, 50,
or 90%) depending on the case investigated. Temperature and
pressure for the reactors are as specied for the NExOCTANE
process while the reactor design paramters (tubes and reactor
length) are modied according to the isobutene conversion.
Plug-ow reactors are known for achieving the highest
conversion per unit volume, and they also require lower
maintenance and less shutdown times. 26 Because the
dimerization reaction is highly exothermic, the reactor is cooled
via a countercurrent stream of cooling water.
An equilibrium-based rigorous two-phase fractionation
model (RADFRAC) was used to model a 14-tray distillation
column; the output stream from the reactor was injected as
feed onto the seventh tray. The adjustment for process
parameters (number of distillation stages, reux ratio, reboiler
duty) was set so as to result in less than 1% iso-butene in the
bottom product and less than 1% of iso-octene in the top
product. The distillation column was operated in the pressure
range of 1015 atm with no pressure drop between the stages
for dierent scenarios of conversion and recycle. The
hydrogenation reactor was modeled by a two-phase continuous
stirred tank reactor operating at 3 MPa and 330 K
(NExOCTANE process conditions specied in ref 13) with
hydrogen injected into the bottom product (isocotene) from
the distillation column.
3573
DOI: 10.1021/ie5032056
Ind. Eng. Chem. Res. 2015, 54, 35703581
Article
A validated three-phase, rate-based, nonequilibrium model

using overall mass-transfer coecients is used for depicting the
CD column performance. This CD model was validated using
our experimental data for the production of isooctane from
isobutene.19,20 Our earlier paper27 presented in detail the
equations including mass, energy, and component balances;
rate expressions; and equilibrium and summation equations for
the CD model. The model used in-house developed binary
mass-transfer coecients for the packing used in the CD
experiments. The model was coded in gPROMS 3.4.3, an
equation-oriented software developed by Process Systems
Enterprise (London, U.K.) for modeling, simulation, optimization, and experimental design studies. Because a rate-based,
nonequilibrium approach28,29 was utilized for modeling the CD
process, gPROMS was preferred over Aspen Plus for the
modeling of the catalytic distillation column. Accommodating
rate expressions for mass transfer to the heterogeneous catalyst
surface adds complexity to the rate expressions. gPROMS oers
the advantage of coding rate expressions involving mass-transfer
coecients and interfacial areas directly into the interface in an
exact form.
Table 2. Comparison of Energy Requirements, Monomer

Utilization, and Carbon Dioxide Emissions with and without
Catalytic Distillation
recycle
0%
25%
50%
75%
100%
0%
25%
50%
75%
100%
0%
25%
50%
75%
100%
100%
recycle
total cooling
(KW/kg
product)
total heating
(KW/kg
product)
nonreacted
isobutene (kg/kg
product)
Conventional Process: 25% Conversion

5 295
5 815
0.75
6 450
7 174
0.5625
8 201
8 123
0.375
9 334
8 869
0.175
11 110
10 719
0
3 795
2 480
0.5
4 319
2 478
0.375
4 988
3 027
0.25
5 304
3 420
0.125
5 605
3 787
0
2 552
944
0.1
2 765
1 142
0.075
4 363
2 210
0.05
4 970
2 614
0.025
5 095
3 221
0
CD Process: Total Conversion
1 520
1 660
0.00004
CO2
produced
(kg/kg
product)
0.895
1.105
1.251
1.366
1.651
0.382
0.381
0.466
0.527
0.583
0.145
0.176
0.340
0.402
0.496
4. RESULTS AND DISCUSSION

This section compares the optimized energy and mass balance
results for the conventional reactor separation process
(conguration A) and the CD process (conguration B).
Various process parameters are compared between the two
congurations that highlight the eciency and advantages of
0.256
Figure 3. CD pilot plant at University of Waterloo for the production of isooctane from isobutene.
3574
DOI: 10.1021/ie5032056
Ind. Eng. Chem. Res. 2015, 54, 35703581
Article
Figure 4. Proles for reactor temperature (left panel) and distillation column (right panel) for conguration A.
Figure 5. Proles of temperatures along CD column (conguration B) for the production of isooctane from isobutene and H2. Nonequilibrium
three-phase gPROMS model (black, vapor phase; red, liquid phase).
at the inlet to about 405 K at 10 cm and then drops to 383 K at

the exit because of counter-current cooling via water at 300 K.
The temperature prole in the distillation column varies from
370 K below the condenser to 410 K above the reboiler.
For conguration B, the gPROMS CD model assumes 75
slices with two reaction zones: a dimerization zone from slice
15 to 30 and a hydrogenation zone between slice 45 and 60.
The hydrocarbon feed (a mixture of 50% isobutene and
isopentane) is injected at slice number 30 while the hydrogen
feed (0.25 mol H2/mol isobutene) is injected at slice 60. The
feed lines in the actual CD column were not heated; therefore,
a steep temperature drop is observed at the two feed injection
points. The temperature prole along the column for vapor
(black) and liquid (red) are shown in Figure 5. It increases
from 365 K at the condenser to 385 K below the dimerization
zone. In the dimerization zone, the temperature suddenly
increases to about 405 K as a result of the exothermic reaction.
Then the temperature suddenly decreases because of injection
of the feed (which was not heated) at slice 30, below the
reaction zone. The temperature continues to rise in the
hydrogenation zone because of the high exothermicity of the
hydrogenation reaction and lower hydrogen feed rates. The
temperature drop at the other feed injection point for
the CD technology over the conventional reactor separation

process.
Because energy integration provided by CD is the primary
focus of investigation, the total energy consumption and
optimized cooling and heating requirements of conguration A
are compared with those of conguration B. Because the CD
process can operate under dierent energy and mass balances
for the same overall product conversion, the optimum
parameters of the recycle ratio and vapor boil-up ratio were
obtained by minimizing the energy needs of the CD column
while maintaining the same octane purity and monomer
separation between the two congurations. A product purity of
0.36 molar fraction of isooctane is the criterion used in the
calculations of energy requirements per unit mass of isooctane
produced for both congurations A and B.
4.1. Proles of Temperature. Figure 4 depicts the
temperature prole along the jacketed water-cooled dimerization reactor and the distillation column for conguration A for
a 50% isobutene conversion case. Temperature proles for
processes are important indicators to study and manipulate the
thermal uxes that enter and exit the system at dierent stages.
The reactor length is 0.5 m. Because it hosts an exothermic
dimerization reaction, the reactor temperature rises from 370 K
3575
DOI: 10.1021/ie5032056
Ind. Eng. Chem. Res. 2015, 54, 35703581
Article
Figure 6. Comparison of the temperature proles, conguration A (blue curve) and conguration B (red curve).
Figure 7. Proles of liquid molar fraction along reactor simulated for conguration A (30% recycle).
process (conguration B). The blue curve is discontinuous

because the process equipment (reactor and distillation
column) are at dierent temperatures. A signicant amount
of energy is wasted in cooling the reaction products, and then
heat is required again to bring them to the separation column
temperature to aid in the separation. This is an area that
signicantly shows the benets brought to the process by
employing CD. The CD process can be assumed to be an
adiabatic reactor because both the reaction and separation units
are housed in one piece of equipment and all of the reaction
heat generated in situ is utilized to aid in the product
separation. This provides substantial savings in the energy for a
CD process, which will be discussed in section 4.3.
hydrogen, just below the hydrogenation zone (slice 60), is not

very steep. Hydrogenation of isooctene is a more exothermic
and faster reaction than the oligomerization of isobutene.
Second, the hydrogen feed rate in the hydrogenation section is
1/4 times that of the feed rate (isobutene and isopentane) in
the oligomerization section in the CD setup. A combination of
a higher, faster heat generation and lower feed rate decreases
the steep temperature drop at the hydrogen injection point.
In Figure 6, the temperature proles with and without CD
are superimposed over each other to eectively provide a
comparison. The blue curve depicts the temperature prole in
the conventional reactor separation process (conguration A)
whereas the red curve depicts the temperature prole in the CD
3576
DOI: 10.1021/ie5032056
Ind. Eng. Chem. Res. 2015, 54, 35703581
Article
Figure 8. Proles of liquid molar fraction along distillation column simulated for conguration A.
Figure 9. Liquid molar fraction proles along CD Column (conguration B).
4.2. Proles of Concentration. The simulated mole

fraction proles along the reactor and the distillation column
(right) are shown in Figures 7 and 8. In the concentration
prole along the reactor length (Figure 7) for a recycle ratio of
30%, the isobutene is consumed and its molar fraction
decreases from 0.25 to 0.03 while isooctene is produced and
its molar fraction increases from 0 to 0.12.
Figure 8 shows the composition proles with respect to
stages along the distillation column in conguration A where
because of the product separation, the isooctene concentration
rises up to 0.43 molar fraction at the last stage. Concentrations
of the volatiles isobutene and isopentane show a decreasing
trend. They have higher concentrations in the upper stages.
Figure 9 shows the composition proles along the CD

column with respect to stage number for conguration B. The
mole fraction of the reacting monomer, isobutene, continuously
drops from the top to the bottom. There is a sharp peak in
mole fraction around stage 30 due to the feed injection (feed
line was not heated) at this stage. Below the condenser the
concentration decreases because of physical separation while in
the dimerization zone, because of the simultaneous reaction
and separation, the slope of the curve is much higher. The mole
fraction of the dimerization product, isooctene, rst increases in
the dimerization zone and then decreases in the hydrogenation
zone. The mole fraction of the nal product, isooctane, is zero
in the zones above the hydrogenation because it is the least
volatile component. Isooctane starts to increase in the
3577
DOI: 10.1021/ie5032056
Ind. Eng. Chem. Res. 2015, 54, 35703581
Article
Figure 10. Heat duty as a function of recycle, CD versus conguration A.
Figure 11. Cold utility requirements as a function of recycle, CD versus conguration A.
component. Under the column operating conditions of 140 psi,

it boils at 403 K. This temperature is breached only in the lower
sections of the column, and as evidenced by the concentration
prole in Figure 9, there is a buildup of isopentane in the lower
sections of the column. In the upper sections of the column,
isopentane exists mostly as a liquid and in trace amounts in
vapor. Isopentane comes out with the reboiler product where
its concentration is almost the same as in the feed and is
removed together with the product stream, completing the
mass balance. The composition of the nal product in the
reboiler is 0.361 molar fraction isooctane, 0.493 molar fraction
isopentane, 0.045 molar fraction isooctene, and less than 0.01
molar fraction isobutene.
4.3. Energy Consumption. Energy eciency for chemical
processes is an area of high interest in an era of rising energy
Table 3. Pounds of CO2 Produced by SteamElectric

Generators for Dierent Fuels30
fuel
coal (bituminous)
coal (subbituminous)
coal (lignite)
natural gas
distillate oil
residual oil
lbs of CO2 per

million Btu
heat rate
(Btu per kWh)
lbs CO2 per

kWh
205
212
10 107
10 107
2.08
2.16
215
117.080
161.386
173.906
10 107
10 416
10 416
10 416
2.18
1.22
1.68
1.81
hydrogenation zone and reaches about 18% above the reboiler.

Reboiler heat increases the composition of isooctane to 0.36
molar fraction in the nal product. Isopentane is the inert
3578
DOI: 10.1021/ie5032056
Ind. Eng. Chem. Res. 2015, 54, 35703581
Article
Figure 12. Carbon dioxide produced in the process as a function of recycle, CD versus conguration A.
2060%, to increase the lifetime of the catalyst and enhance

selectivity of dimer by reducing the formation of the
byproducts, i.e., trimers and higher oligomers. Our simulation
results demonstrate CD can provide signicant reduction of hot
and cold utilities for the isobutene dimerization process.
Moreover, Table 2 also depicts that CD leads to signicantly
better utilization of the monomer isobutene, as an added
benet for the isooctane process. Figures 10 and 11 relate the
signicant energy integration CD brings into the isooctane
production process via reduction in utility requirements.
4.4. Reduction in Carbon Dioxide Emissions. While the
focus of this research is on energy savings provide by CD, this
study also provides quantitative estimates translating energy
savings into reduction of carbon dioxide (CO2) and GHG
(greenhouse gas) emissions. Reduction of CO2 and GHG
emissions for a process provides both economic and environmental merits. As an economic factor, reduction in CO2 and
GHG emissions changes the evaluation of energy eciency
improvements (as determined by environmental regulations
and market economy) and also provides viability and
sustainability for projects that otherwise would not be
economical.
Table 2 compares the energy requirements in CD
conguration (kilowatt per kilogram of isooctane product)
for the isooctane process against various process parameters for
conguration A. Results demonstrate signicant cooling water
savings (up to 7 times) at varying recycle ratios and signicant
cooling and hot water duties at higher recycle ratios (up to 11
times) in the CD conguration. The challenge is to relate these
energy savings into GHG savings by nding a suitable
conversion ratio. Figures for CO2 reductions are achieved via
emission factors provided by the U.S. Energy Information
Administration (EIA),30 that compute the amount of CO2
produced per kilowatt hour (kWh) for specic fuels and specic
types of generators. These indicators are listed in Table 3.
Most distillation columns in renery operations have watercooled condensers using cooling water that is inexpensive.31
Cooling operations in the condenser are normally achieved via
large volumes of water. This practice essentially does not
Table 4. Eect of Addition of Hydrogenation Reaction on

the Energy and Mass Balance of CD Column
CD model
dimerization only
dimerization plus
hydrogenation
feed (mol/s)
isooctene: 0.000 25
isopentane: 0.000 27
isooctene: 0.000 25
reboiler
duty (W)
condenser
load (W)
14.95
13.04
13.3
12.1
isopentane: 0.000 27
hydrogen: 0.000 063
costs and environmental concerns. The most noteworthy

benet that CD brings to the isooctane production process is in
terms of signicant reductions in utility requirements. Table 2
lists the total energy consumption per kilogram of isooctane
produced with and without CD at dierent recycle rates and
dierent conversions. Comparisons are made per unit mass of
product (per kilogram of isooctane) at the same purity (0.36
molar fraction iso-octane), so varying ow rates between the
two congurations ceases to be a factor. The conventional
process (conguration A) is modeled at varying conversions
and reux ratios, whereas the CD process (conguration B) is
modeled at total reux. Results demonstrate that at all
conversions and recycle rates, the CD process requires
signicantly less cold utilities (up to 7 times savings in cold
utilities). This is expected because there is massive cooling
water consumption in a conventional reactor separation process
to cool the reactor and to protect the catalyst from deactivation
and products from the exothermic reaction heat before they
enter the separation units. In a CD process, the in situ heat
generated from the exothermic reactions is dissipated from the
reaction zones toward separation of products so that minimal
cooling energy is required. In the case of hot utility
requirements, CD outperforms conguration A at all
conversion and recycle scenarios except at very high isobutene
conversions of 90% and low recycle rates. It should be noted a
90% conversion of isobutene is not feasible in an industrial
reactor conguration and that conversions in isobutene
dimerization processes are typically kept low, in the range of
3579
DOI: 10.1021/ie5032056
Ind. Eng. Chem. Res. 2015, 54, 35703581
ACKNOWLEDGMENTS
Financial support from the Natural Sciences and Engineering
Research Council of Canada is gratefully acknowledged. A.G.
acknowledges a Trillium scholarship for graduate studies
provided by the Ontario Government, Canada.
contribute to CO2 emissions. Hence, only the hot utilities were

scaled by the conversion factor in calculating the CO2
emissions. Natural gas is the preferred fuel for use in petroleum
reneries utility systems;32,33 hence, the conversion ratio for
natural gas (1.22) was used.30 Because we are directly scaling
the hot utility requirements by an emission factor, the relative
savings in carbon dioxide emissions is the same as the hot utility
requirements (higher savings in carbon dioxide emissions at
higher recycle ratios (up to 11 times)). Results for carbon
dioxide emissions in terms of kilograms per kilogram of
isooctane are tabulated in the last column of Table 2 and also
shown in Figure 12.
As discussed earlier, most conventional reactor separation
congurations run at conversions between 20% and 60% per
pass. Assuming a conventional process operating at 50%
conversion and 100% recycle, a reduction of 0.327 kg of carbon
dioxide emissions is achieved per kilogram of isooctane
product. Considering an annual production of 520,000 tons
of isooctane product from a NExOCTANE process,34 this
translates to a reduction of 170,000 tons of carbon dioxide
emissions per year.
4.5. Eect of Reaction Heat on the Reboiler and
Condenser Loads. Results in section 4.3 demonstrated a
signicant reduction in reboiler duty in the CD operation for an
exothermic process. To further explore the eect of the
hydrogenation reaction heat on the energy balance in the CD
process, two CD models were compared: a CD process with
dimerization only and a CD model with two reactions, namely,
dimerization and hydrogenation. When the hydrogenation
reaction was carried out in the same CD column for
dimerization, a 20% decrease in the reboiler and condenser
duties was observed, as listed in Table 4. These results are
expected because of the exothermicity of the hydrogenation
reaction.
REFERENCES
(1) Srivastava, S. P.; Hancsok, J. Fuels from Crude Oil (Petroleum).

In Fuels and Fuel-Additives; John Wiley & Sons, Inc.: Hoboken, NJ,
2014; pp 48120.
(2) Krivan, E.; Marsi, G.; Hancsok, J. Investigation of the
Oligomerization of light olefins on ion exchange resin catalyst. Hung.
J. Ind. Chem. 2010, No. 38, 5357.
(3) Coelho, A.; et al. 1-Butene oligomerization over ZSM-5 zeolite:
Part 1 Effect of reaction conditions. Fuel 2013, 111 (0), 449460.
(4) Bellussi, G.; et al. Oligomerization of olefins from Light Cracking
Naphtha over zeolite-based catalyst for the production of high quality
diesel fuel. Microporous Mesoporous Mater. 2012, 164 (0), 127134.
(5) Ouni, T.; et al. Isobutene dimerisation in a miniplant-scale
reactor. Chem. Eng. Process. 2006, 45 (5), 329339.
(6) Kamath, R. S.; et al. Comparison of Reactive Distillation with
Process Alternatives for the Isobutene Dimerization Reaction. Ind. Eng.
Chem. Res. 2006, 45 (8), 27072714.
(7) NeXoctane Process KBR. http://www.kbr.com/Technologies/
Process-Technologies/NExOCTANE/ (accessed July 7, 2011).
(8) Ward, T. Aerospace Propulsion Systems; John Wiley & Sons (Asia)
Pte Ltd.: Singapore, 2010:.
(9) Totten, G. E.; Westbrook, S. R.; Shah, R. J. Fuels and Lubricants
Handbook - Technology, Properties, Performance, and Testing; ASTM
International: West Conshohocken, PA, 2003; MNL 37WCD.
(10) Bowman, W. G. H.; Stadig, W. P.. Dimerization of isobutene.
U.S. Patent 4,100,220, July 11, 1978
(11) Kamath, R. S.; et al. Process Analysis for Dimerization of
Isobutene by Reactive Distillation. Ind. Eng. Chem. Res. 2006, 45 (5),
15751582.
(12) Centi, G., Siglinda, P., Ferruccio, T. Sustainable Industrial
Chemistry: Principles, Tools and Industrial Examples; Wiley-VCH:
Weinheim, Germany, 2009.
(13) Birkho, R. N.; Nurminen, M. NExOCTANE Technology for
Isooctane Production. In Handbook of Petroleum Rening Processes, 3rd
ed.; Meyers, R., Ed.; McGraw-Hill: New York, 2004.
(14) Kolah, A.; Zhiwen, Q.; Mahajani, S. Dimerized isobutene: An
alternative to MTBE. In Chem. Innovation; ACS: Washington, DC,
2001; pp 1521.
(15) Alkylation vs. Dimerization: What are the limitations of modern
dimerization technology? NPRA Annual Meeting, STRATCO, Spring
2000.
(16) Dow Chemicals. DOW Ion Exchange Resins - Catalysts for
Dimerization of Isobutylene - Iso Octane. http://dowac.custhelp.com/
app/answers/detail/a_id/16116/~/dow-ion-exchange-resins--catalysts-for-dimerization-of-isobutylene---iso-octane (accessed May
27, 2014), 2013.
(17) Ng, F. T. T.; Rempel, G. L. Catalytic Distillation. In Encyclopedia
of Catalysis; John Wiley & Sons, Inc.: Hoboken, NJ, 2002.
(18) Kim, S.; Douglas, P. L. Optimisation of Methyl Tert-Butyl-Ether
(MTBE) Synthesis Processes using Aspen-Plus. Dev. Chem. Eng. Miner.
Process. 2002, 10 (12), 89103.
(19) Lei, Z.; et al. Catalytic distillation for the synthesis of tert-butyl
alcohol with structured catalytic packing. Catal. Today 2009, 147
(Supplement), S352S356.
(20) Ming, J., et al. The One-Step Synthesis of MIBK via Catalytic
Distillation. In Catalysis of Organic Reactions; CRC Press: Boca Raton,
FL, 2006.
(21) Lei, Z.; et al. Synthesis of dimethyl ether (DME) by catalytic
distillation. Chem. Eng. Sci. 2011, 66 (14), 31953203.
(22) Mings, W. J. Energy Ecient Renery Process Developed with
U.S. D.O.E. Support. In Proceedings from the Fifth Industrial Energy
Technology Conference, Volume II, Houston, TX, April 1720, 1983.
5. CONCLUSION
The eciency of a green reactor engineering technology, CD,
in terms of energy integration and material savings is compared
to a conventional reactor and distillation process for the
industrial production of isooctane from isobutene. Results
establish the eciency of the CD technology over the
conventional reactor separation technology in terms of energy
requirements and isobutene feed utilization. The CD model
predicts signicant cold and hot utility savings (up to 711
times) and reduction of carbon dioxide emissions (up to 11
times) at dierent ratios of conversion and recycle. The results
provide evidence that CD is a very attractive process
intensication technique which brings substantial energy
savings, more ecient isobutene feed utilization, and reduction
of greenhouse gas emissions.
Article
AUTHOR INFORMATION
Corresponding Author
*E-mail: fttng@uwaterloo.ca.
Present Address
B.M.G: Faculty of Advanced Sciences and Technologies,

University of Isfahan, Iran.
Notes
The authors declare no competing nancial interest.

3580
DOI: 10.1021/ie5032056
Ind. Eng. Chem. Res. 2015, 54, 35703581
Article
(23) Gaurav, A.; et al. Transesterification of Triglyceride to Fatty

Acid Alkyl Esters (Biodiesel): Comparison of Utility Requirements
and Capital Costs between Reaction Separation and Catalytic
Distillation Configurations. Energy Fuels 2013, 27 (11), 68476857.
(24) Sarkar, A. A Catalytic Distillation Process for One-step Production
of Isooctane from IsoButeneProcess Development, Modelling and
Analysis. Ph.D. Thesis, University of Waterloo, Waterloo, ON, Canada,
2005; p 364.
(25) Xu, Y. Numerical Modelling and Experimental Study of Catalytic
Distillation for Olen Oligomerization and Hydrogenation. Ph.D. Thesis,
University of Waterloo, Waterloo, ON, Canada, 2006.
(26) Fogler, H. S. Elements of Chemical Reaction Engineering, 4th ed.;
Prentice Hall Professional Technical Reference: Upper Saddle River,
NJ, 2006.
(27) Zheng, Y.; Ng, F. T. T.; Rempel, G. L. Catalytic Distillation: A
Three-Phase Nonequilibrium Model for the Simulation of the Aldol
Condensation of Acetone. Ind. Eng. Chem. Res. 2001, 40 (23), 5342
5349.
(28) Krishnamurthy, R.; Taylor, R. A nonequilibrium stage model of
multicomponent separation processes. Part I: Model description and
method of solution. AIChE J. 1985, 31 (3), 449456.
(29) Xu, Y.; et al. A three-phase nonequilibrium dynamic model for
catalytic distillation. Chem. Eng. Sci. 2005, 60 (20), 56375647.
(30) U.S. Energy Information Administration. How much carbon
dioxide is produced per kilowatthour when generating electricity with fossil
fuels? http://www.eia.gov/tools/faqs/faq.cfm?id=74&t=11 (accessed
June 16, 2014).
(31) Otts, L. E., Jr. Water Requirements of the Petroleum Rening
Industry; 1330-G; United States Department of the Interior, United
States Government Printing Oce: Washington, DC, 1963.
(32) U.S. Energy Information Administration. Renery Capacity
Report, 2013.
(33) Taraphdar, T.; Yadav, P.; Prasad, M. Natural gas fuels the
integration of rening and petrochemicals. Digital Rening 2012
(34) NesteOil. Neste Oils iso-octane plant sale conrmed. http://
www.nesteoil.com/default.asp?path=1,41,540,1259,1260,18523,18566
(accessed July 15, 2014), 2012.
3581
DOI: 10.1021/ie5032056
Ind. Eng. Chem. Res. 2015, 54, 35703581
Biodegradable Nanocomposites of Cellulose Acetate

Phthalate and Chitosan Reinforced with Functionalized
Nanoclay: Mechanical, Thermal, and Biodegradability
Studies
Aashish Gaurav,1 A. Ashamol,2 M. V. Deepthi,2 R. R. N. Sailaja2
1
Deparment
2
of Chemical Engineering, IIT, Kharagpur, West Bengal, India

The Energy and Resources Institute, Bangalore 560071, Karnataka, India
Received 16 December 2010; accepted 20 July 2011

DOI 10.1002/app.35591
Published online 19 December 2011 in Wiley Online Library (wileyonlinelibrary.com).
ABSTRACT: Biodegradable nanocomposites of cellulose
acetate phthalate and chitosan reinforced with functionalized nanoclay (NC) were prepared. The NC loading was
varied from 0 to 10%. The mechanical and thermal properties have been investigated for these composites. The
nanocomposites exhibited enhanced mechanical properties due to the addition of NC. The scanning electron
micrographs of the blend specimens also support the
above observations. Thermogravimetric analyses were
carried out to assess the degradation stability of the

blends. The blend shows an increase in the rate of biodegradation and water uptake with higher loading of NC.
The exfoliation of NC was analyzed by X-ray diffraction
C 2011 Wiley Periodicals, Inc. J Appl Polym Sci 125:
studies. V
INTRODUCTION
been investigated. It has been found earlier that interactions do exist between chitosan and cellulose as
reported by Hasegawa et al.5 An interpenetrating
polymer network forms between cellulose and chitosan with decreasing crystallinity as chitosan loading
is increased as reported by Cai and Kim6 An
increased thermal stability of chitosan-montmorillonite biocomposite films dependent of clay content was
reported by Altinisik et al.7 Miscibility studies of chitosan blend with cellulose ethers revealed that the
blends are partially miscible in dry state although,
hydrogen bonding exists between the functionalized
groups.8 However, cellulose acetate/chitosan blend
films have been found to have improved miscibility
and good mechanical properties as studied by Liu
and Bai.9 Similar studies were also reported by Shih
et al.10 Clay reinforced cellulose acetate nanocomposites show an improvement in mechanical properties
and thermal stability.11 In this article, the effect of
adding NC to a blend of cellulose acetate phthalate
(CAP) and chitosan has been investigated. The NC
used is surface modified to enhance dispersion and
bonding with the blend components.
The development of high performance bioplastics is

gaining prominence owing to increasing plastic pollution and also to conserve the diminishing global petroleum reserves. Further, biopolymers are inexpensive,
renewable, and sustainable alternatives, which can
replace petrochemical-derived synthetic polymers.
Cellulose and chitosan are among the most abundant
natural biopolymers, which are inexpensive, renewable, and biodegradable with antibacterial properties.
Biobased nanocomposites are produced in which
atleast one component is nanosized and acts as a
reinforcement even with low content.1 Thus, chitosan
reinforced with nanosized cellulose whiskers, which
have been reported to exhibit improved tensile and
water resistance.2 El-Tahlawy et al.3 developed spinnable esterified chitosan butyrate/cellulose acetate
blend fibers with enhanced properties. Films of iminochitosan with cellulose acetate yielded smooth homogeneous films4 with good mechanical strength.
In this study, an esterified cellulose derivative has
been blended with chitosan and reinforced with surface functionalized nanoclay (NC). The mechanical,
thermal, and biodegradability characteristics have
Correspondence to: R. R. N. Sailaja (rrnsb19@rediffmail.
com).
Journal of Applied Polymer Science, Vol. 125, E16E26 (2012)
C 2011 Wiley Periodicals, Inc.

V
E16E26, 2012
Key words: cellulose acetate phthalate; chitosan; nanoclay;

mechanical and thermal properties; biodegradability
EXPERIMENTAL
Materials
CAP (degree of substitution for acetyl and phthalyl
groups are 1.07 and 0.77, respectively) with
STUDY OF FUNCTIONALIZED NANOCLAY
molecular weight 2534.12 was purchased from GM

Chemicals, Mumbai. Chitosan (with 85% deacetylation and molecular weight ranging from 10,000 to
15,ooo) was purchased from Marine chemicals, Cochin, Kerala. Silane-treated NC was obtained from
Sigma Aldrich (USA). Glycerol and other common
solvents were obtained from S.D. Fine Chem,
Mumbai.
Preparation of blend
A total of 100 g mixture of CAP (60 g) and chitosan
(40 g) powders were taken for the preparation of
composites. NC quantity was varied from 0 to 10 wt
%. The mixture was mixed in a kitchen mixer for 10
min. Glycerol (40% of the weight of the mixture)
was added to the mixture and then sonicated using
Ultra sonicator (Branson, 2510E/DTH) for 30 min.
Water (100 mL) was then added to the mixture and
the contents were kneaded to form a dough. The
blend of pure CAP and chitosan without NC also
contained the same amount of glycerol and water.
The dough samples were kept in zip-locked plastic
packets in refrigerator for further processing.
Compression molding
The dough was partially dried prior to compression
molding. The semi-dried blend was placed in a
mold covered with two polished stainless steel
plates and then compression molded using a locally
fabricated hot press. Sheets were molded at 130 C
under a pressure of 15 MPa for 3 min, and then
cooled to about 50 C for 15 min under constant pressure before releasing the pressure for demolding.
The sheets were then cut into rectangular strips and
these strips were subjected to mechanical testing.
Mechanical properties of the blend
Tensile properties
The tensile properties of the blends were measured
by Zwick UTM (Zwick Roell, ZHU, 2.5) with Instron
tensile flat surface grips at a crosshead speed of 2
mm/min. The tensile tests were performed as per
ASTM D 638 method. The specimens tested were of
rectangular shape having length, width, and thickness of 7, 1.5, and 0.3 cm, respectively. A minimum
of five specimens were tested for each variation in
composition of the blend and results were averaged.
Predictive theoretical models have been used to analyze the observed experimental results.
Flexural properties
The flexural properties of the blends were measured
by Zwick UTM (Zwick Roell, ZHU, 2.5) with a pre-
E17
load speed of 10 mm/min. The tests were performed

as per ASTM D 790-03 method. The samples were
having a length of 5 cm, width of 2 cm, and a thickness of 0.3 cm. A minimum of five specimens were
tested for each blend and the results were averaged.
Compressive properties
The compressive properties of pure CAP, pure chitosan, and the blends were performed as per ASTM D
695 by Zwick UTM (Zwick Roell, ZHU, 2.5) with a
preload of 4.5 kN and a test speed of 3 mm/min.
The samples were having a length of 3 cm, width of
2 cm, and a thickness of 0.3 cm. A minimum of five
specimens were tested for each composition and the
results were averaged.
Thermal analysis
Thermogravimetric analysis (TGA) was carried out
for the blends using Perkin-Elmer Pyris Diamond
6000 analyzer in nitrogen atmosphere. The samples
were subjected to a heating rate of 100C/min in the
heating range of 40600 C using Al2O3 as the reference material.
Differential scanning calorimetry (DSC) of the
nanocomposite specimens have been performed in a
Mettler Toledo instrument (model DSC 822e; Mettler
Toledo AG, Switzerland). Samples are placed in
sealed aluminum cells, with a quantity of less than
10 mg and scanning at a heating rate of 10 C/min
up to 250 C.
FTIR spectroscopy
Fourier transform infrared spectroscopy (FTIR; Perkin-Elmer spectrum 1000) analysis for the Pure CAP,
chitosan, and the nanocomposites was performed.
X-ray diffraction
XRD measurements for the nanocomposites have
been performed using advanced diffractometer
(PANalytical, XPERT-PRO) equipped with a Cu-ka
radiation source (X 0.154 mm). The diffraction
data were collected in the 2y range of 330 using a
fixed-time mode with a step interval of 0.05 .
Blend morphology
Scanning electron microscopy (SEM; LEICA.S440,
Model 7060) is used to study the morphology of the
fractured and unfractured specimens. The specimens
are gold sputtered prior to microscopy. The SEM
morphology of the unfractured blend specimens was
taken after soaking the samples in dilute sulfuric
Journal of Applied Polymer Science DOI 10.1002/app
E18
GAURAV ET AL.
RESULTS AND DISCUSSIONS

Bionanocomposites of CAP and chitosan have been
prepared and characterized using FTIR, XRD, and
SEM. The mechanical and thermal properties of the
nanocomposites have also been studied.
FTIR spectroscopy
Figure 1 FTIR Spectra of pure cellulose acetate phthalate,

chitosan, and silanated nanoclay. [Color figure can be
viewed in the online issue, which is available at
wileyonlinelibrary.com.]
acid for 24 h and then drying in air after thoroughly

rinsing it in distilled water.
Water absorption
Water absorption index of the samples were measured according to ASTM D 570-81 with minor modification.12 The dried sample is weighed and submerged in distilled water at room temperature for
24 h. The extra water on the surface of the specimen
after soaking is removed by placing it in an air oven
at 50 C and the specimen were weighed again. The
container without the soaking specimen is placed in
an air oven at 50 C for 72 h to evaporate the water,
and the water-soluble content obtained was equal to
the increase in container weight. The absorption
(AB) is then calculated by the following eq. (1):
AB W1 Wo Wsol =Wo
Figure 1 shows the FTIR spectroscopy of neat chitosan, CAP, and NC, while that of CAPchitosan nanocomposites are shown in the Figure 2. Spectroscopy
of neat chitosan, CAP, and NC are given for the sake
of comparison. The blends with 0% NC does not
have the peak at 1643 cm1, which is a characteristic
of amide I band. This is mainly attributed to the fact
that the amide group of chitosan has reacted with the
carboxyl group of CAP. Silanated-NC has two main
peaks. The first one is at 1025 cm1 14 for the SiOstretching of silicate present and also of the interaction with platelet surface. The second peak is at 1603
cm1 for ANH2 (primary amine) stretching, which
does not appear for the blends. The characteristic
bands for CAP at 1035, 1239, 1589, 1724, and 2913
cm1 are respectively for CO stretching, COC
stretching, CC conjugated vinyl aromatic ring,
CO carboxyl group, and asymmetric and symmetric
stretching of methyl CH groups.15 All the above
bands are also seen in the blends. However, the other
bands overlap with that of CAP and chitosan.
Stressstrain curves
The engineering stressstrain curves for CAPchitosan nanoblends are shown in Figure 3. The stress
strain curves for neat CAP and chitosan are also
given in the figure for comparison. CAP [curve (a)] is
(1)
where W1, Wo, and Wsol are the weight of the specimen containing water, the weight of the dried specimen, and the weight of the water-soluble residues,
respectively.
Biodegradation
The biodegradation of the blend specimens were carried out by soil burial method.13 Soil-based compost
was taken in small chambers. Humidity of the chambers was maintained at 4045% by sprinkling water.
The chamber were stored at 3035 C. Rectangular
specimens were buried completely into the wet soil
at a depth of 10 cm. Samples were removed from the
soil at constant time intervals (15 days) and washed
gently with distilled water and dried in vacuum oven
at 50 C to constant weight. Weight loss percentages
of the samples with respect to time were recorded to
determine the extent of biodegradation.
Figure 2 FTIR Spectra of CAPchitosannanoclay blends.

[Color figure can be viewed in the online issue, which is
available at wileyonlinelibrary.com.]
E19
blend particles. CAP is relatively more ductile than

chitosan and NC, and this also helps in further
anchoring the two blend components.
Figure 3 Plots of engineering stressstrain curves for

pure cellulose acetate phthalate, chitosan, and CAPchitosannanoclay blends.
more ductile when compared with chitosan [curve

(b)], which has poor stress resistance and has brittle
characteristics. The blend, however [curve (c)], exhibits higher stress value when compared with either of
the blend components. Addition of NC [curve (d)(f)]
improves both stress as well as strain values owing
to its reinforcing effect among the blend components.
The optimal stress values at 6% NC [curve (g)]
shows, values higher than either CAP or chitosan
with a strain values higher than that of chitosan.
However, higher loading of NC of 10% [curve (h)]
has a detrimental effect on the mechanical properties
of the blend. This may be due to saturation of reactive sites of CAP or chitosan, which can react with
the amine group of silane-treated NC. Thus, excess
NC behaves like a separate third phase, which
reduces the properties.
Effect of NC addition on mechanical properties
Tensile properties
Figure 4 shows the plot of relative tensile properties
(i.e., relative to blend without NC) versus volume
fraction of NC (/). The relative elongation at break
(REB) increases by adding NC up to / 1.0147 (i.e.,
6% NC) and reaches an optimal value of 1.8. This is
contrary to the expectation that the addition of rigid
NC particles had to lower the REB values and, in
this case, the elongation at break values are even
higher than either CAP or chitosan. A similar observation was made by Balakrishnan et al.16. Sue
et al.17 also did not observe a lowering in strain
by adding rigid zirconium phosphate particles,
although the mechanism is not well understood. It
may be due to the fact that the amine group of silane-treated NC gets coated/interacted with CAP
and the plastic deformation gets initiated around the
Figure 4 Plots of relative tensile properties versus volume fraction of nanoclay. (a) Relative tensile strength, (b)
relative Youngs modulus, (c) relative elongation at break.
[Color figure can be viewed in the online issue, which is
available at wileyonlinelibrary.com.]
E20
GAURAV ET AL.
The relative tensile strength (RTS) of the nanocomposites also showed a 10% increase when compared
with blends without NC. The optimal value is
reached at / 1.0147 (i.e., 6% NC loading) beyond,
which it is detrimental for the nanoblends. It may be
due to the fact that all the reactive sites have been
used up to this NC loading and with further NC
loading, they agglomerate due to the presence of
excess unreacted sites.
The relative tensile modulus (RYM) values also
reduce due to the addition of NC and the curve is a
mirror image of that obtained for REB, although in
most cases, the addition of rigid particles such as
starch to a ductile matrix exhibits an increase in
modulus.18
To further analyze the obtained experimental
results, the following predictive theories have been
used as described below.
Figure 4(a) shows the plot of RTS values versus
volume fraction of NC (/). The volume fraction of
NC, was calculated using the following eq. (2):
wi =qi
/i P
wi =qi
(2)
In eq. (2) wi and qi is the weight fraction and density of component i in the blend. The density values
of CAP, chitosan, and NC have been measured to be
0.92, 0.54, and 0.36 g/cm3, respectively.
Three models were used to compare the obtained
experimental tensile strength values. The first is the
Nicolais and Narkis model,19 which is as follows.
RTS
rb
1 1:21/2=3
ro
(3)
In eq. (3), rb and rO are the tensile strength of the

nanoblends and tensile strength of the blend without
NC reinforcement. The model assumes that addition
of filler reduces the effective cross-sectional area and
there exists no adhesion between filler and matrix.
As observed in plot 4(a), the theoretical values of eq.
(3) and the experimental RTS values does not match.
The second model is the HalpinTsai model,19
which is given below in eq. (4)
RTS
rb 1 GgT /
ro
1 gT /
(4)
In eq. (4), gT is given by the following equation.

gT
RT 1
7 5t
and G
RT G
8 10t
(5)
In eq. (5), RT is the ratio of tensile strength of NC

to the tensile strength of CAP/chitosan blend without NC. RT was determined by trial and error by
minimizing the difference between the obtained experimental results and calculated theoretical values,
this was found to be 0.548. t is the Poissons ratio of

CAP/chitosan blend, which is taken to be 0.37.20
The predicted values obtained from eq. (5) are closer
to the experimental values when compared with that
obtained by NicolaisNarkis model, as the Halpin
Tsai model assumes good adhesion between the
blend components.
The third model is the Turcsanyi model,21 which
includes an interfacial parameter B, which is a measure of extent of adhesion of the filler with the matrix and the equation, is given below as follows.
RTS
rb
1/
expB/
ro 1 2:5/
(6)
The value of B was determined to match with the

experimental results and this was found to be 2.4,
which indicates good adhesion. The amine group of
silane on NC interacts well with the blend components and thus leads to enhanced tensile strength
values. A similar analysis using the above model
was carried out by Zou et al.22 for polyesteramide
composites with different fillers. Thus, blends with
poor tensile strength showed a B value of 0.25 (no
adhesion), while higher values (e.g., 3.44 for talc)
indicated better adhesion. The values obtained for
eq. (6) are also shown in Figure 4(a). The predicted
values of the Turcsanyi model are higher than that
obtained for eq. (3), but theoretical values obtained
from HalpinTsai model are closer to the experimental data.
Figure 4(b) shows a plot of relative tensile modulus (RYM) versus volume fraction of NC. The RYM
values reduce with the increase in NC content as
described earlier.
Three models have been used to analyze the
obtained experimental data. The first is the Kerners
model,19 which assumes no interaction between the
blend components and is given by:

Eb
/
151 t
RYM 1
Eo
1 / 8 10t
(7)
In eq. (7), Eb and EO are the tensile modulus of the

nanoblends and that of the CAPchitosan blend
without NC, respectively.
The theoretical values obtained from eq. (7) are
also plotted in Figure 4(b). The experimental data
does not match with the predicted values indicating
the existence of interaction between the blend components. The model for improved matrixfiller interaction is described by the HalpinTsai model given
below in eq. (8).
RYM
1 gm /
1 gm /
(8)
E21
where gm is given by:

gm
Rm 1
Rm G
(9)
In eq. (9), Rm is the ratio of filler modulus to matrix modulus. Rm was determined by trial and error
to match with the experimental results as described
earlier and was found to be 0.004. The values determined using eq. (8) is also plotted in Figure 4(b).
The predicted values are closer to the experimental
data and the trend also matches with that obtained
experimental results.
The third model is the one developed by Sato and
Furukawa,21 which includes an adhesion parameter
n, which varies from 0 to 1 for perfect adhesion to
no adhesion. The model is described by eq. (10)
below.
2
RYM 4 1
2=3
2 2/1=3
!
1 wn
wf 5

1/
/
2=3
1=3
(10)
where,
!

/
1 /1=3 /2=3
w
3
1 /1=3 /2=3
(11)
The value of n was determined to match with the

experimental results and has been found to be 0.7.
The value of n indicates good adhesion and is
between the two extremes. The theoretical data predicted using eq. (10) show a trend similar to that
experimentally observed as shown in Figure 4(b).
Figure 4(c) shows a plot of REB versus volume
fraction (/) of NC. As observed earlier, there is a
significant improvement in REB values by the addition of NC. Nielsens model23 has been used to analyze the observed values. The equation for this
model is given below.
REB

2=3
2b
1 k/
2o
(12)
In eq. (12), eb and eO is the elongation at break for

the nanoblends and that of the blend without NC,
respectively. In eq. (12), k is an adjustable parameter,
which depends on filler geometry. The value of k
was computed by trial and error to get the best
match for the obtained experimental results and this
was found to be 0.09. The theoretical values plotted
in Figure 4(c) do not match with the experimental
values, which indicate that this model cannot
explain the observed trend.
Figure 5 Variation of relative compressive properties

with percentage nanoclay. (a) Relative compressive
strength, (b) relative compressive modulus. [Color figure
can be viewed in the online issue, which is available at
Compressive properties
Figure 5(a) shows a plot of relative compressive
strength (RCS) of the blends versus percentage NC
loading. The RCS value reaches an optimal value at
6% NC loading and the compressive strength of the
blend increases by 22% (RCS 1.22) when compared with blend without NC. The compressive
strength of the blends reduce due to the addition of
chitosan as it is brittle compared with CAP (the
compressive strength of CAP is 12.96 Mpa). The
nonlinearity of compressive properties has been discussed by Siqueira et al.24 A threefold increase in
compressive properties for biocompatible nanocomposites has been observed by Shi et al.25
Figure 5(b) shows the plot of relative compressive
modulus (RCM) versus percentage NC loading for
CAPchitosan blends. The RCM values show a
decreasing trend with the addition of functionalized
E22
GAURAV ET AL.
NC. The RCM values slightly reduce from 1.0 (without NC) to 0.947 with 6% NC as the optimal value.
The plasticizing effect of ester group also plays a
role in the reduction of modulus values as esters
behave like internal plasticizers.26
Flexural properties
Figure 6(a) shows the relative flexural strength (RFS)
for the blends versus percentage NC loading. Addition of NC to the blend did not show any improvement of RFS values. The RFS value for blend with 3,
6, and 8% NC were, respectively, found to be 0.966,
0.95, and 0.996.
Figure 6(b) show the relative flexural modulus
(RFM) versus %NC loading. The blend containing
no NC has a RFM-value of 1.0, while an optimal
value of 1.014 was observed with 6% NC. Thus, in
Figure 6 Variation of relative flexural properties with

percentage nanoclay. (a) Relative flexural strength, (b) relative flexural modulus. [Color figure can be viewed in the
online issue, which is available at wileyonlinelibrary.com.]
Figure 7 TGA themograms of pure CAP, chitosan, CAP

chitosannanoclay blends.
Figure 8 DSC thermograms for pure CAP, chitosan,

CAPchitosannanoclay blends. [Color figure can be
viewed in the online issue, which is available at
E23
degradation takes place29 and at 376 C, 99% weight

loss occurs. Chitosan [curve (b)] undergoes a single
stage degradation at 290.5 C (with 54.7% weight
loss) owing to the degradation and deacetylation of
chitosan.30
Among the CAP and chitosan, the latter has a better thermal stability probably owing to inter- and
intramolecular hydrogen bonding. Similar observations have been reported by various authors.3133
The blends of CAP and chitosan [curve (c)] exhibit
better interaction with only 47% weight loss at
310 C. Addition of NC slightly enhanced the thermal stability and an increase in char content was
observed [curves (d) and (e)].
DSC thermograms
The DSC thermograms for CAPchitosan blends are
shown in Figure 8(a). Pure CAP shows a Tg at
Figure 9 XRD spectra of pure CAP, chitosan, CAPchitosannanoclay blends. [Color figure can be viewed in the
online issue, which is available at wileyonlinelibrary.com.]
general, the flexural properties did not exhibit any

improvement on NC addition. Sorrentino et al.27
suggested that the enhancement of flexural properties is mainly due to the formation of three dimensional network of silicate layers. Similar observation
for soy-based nanocomposites has been reported by
Sithique et al.28
Thermogravimetric analysis
Figure 7 shows the TGA thermograms of CAPchitosan blends. The thermograms for neat CAP, chitosan, and NC are also included in the figure for the
sake of comparison.
Neat CAP [curve (a)] shows a main degradation
peak at 310 C (with 78.1% weight loss) with a
shoulder peak at 249 C (with 40.7% weight loss). At
this temperature, elimination of acetyl and phthalyl
groups takes place and the onset of backbone chain
Figure 10 Scanning electron micrographs showing blend

morphology for CAPchitosanNC. (a) CAPchitosan
blend without NC, (b) CAPchitosan blend with 10% NC.
E24
GAURAV ET AL.
Figure 11 Scanning electron micrographs showing tensile fracture morphology for CAPchitosanNC. (a) Neat chitosan,
(b) neat CAP, (c) CAPchitosan blend without NC, (d) CAPchitosan blend with 6% NC, (e) CAPchitosan blend with
10% NC.
146 C. Similar observations have been reported by

Rao et al.34,35 The Tg of pure chitosan is at 142 C,
which agrees with the results of Dong et al.36
Another transition at 228 C may be attributed to liquidliquid transition.36 The composites without NC
(0% NC) show a Tg at 140 C. Addition of 2 and 8%

NC shows a Tg at 142.2 and 143 C, respectively.
Thus, the Tg values for the blends are between the
two extremes, i.e., between that of neat CAP and
neat chitosan.
X-ray diffraction
The XRD patterns of CAPchitosan biocomposites
are shown in Figure 9. This figure also includes the
XRD profiles for neat NC, CAP, and chitosan.
Neat NC indicates a characteristic diffraction at 2y
values of 8.03 , 19.778 , 24.73 , and 26.63 . Neat CAP
has a main broad peak at a 2y value of 22.038 ,
while that for chitosan the crystalline peak is at
20.792 . The blend of CAP and chitosan (without
NC) has a peak at a 2y value of 19.5574 accompanied by a small shoulder peak at 22 . The curve for
blend loaded with 3% NC and 6% NC is also shown
in Figure 9. The characteristic peak at 8.03 of NC is
missing indicating that NC has formed an exfoliated
structure with the blend components. Similar observation has been made for chitosan-based nanocomposites as reported by Wang and Wang.37
Blend morphology
Figure 10 shows the blend morphology of CAPchitosan blend. The samples were etched in acid solution
for 3 h so that the chitosan phase is removed. Figure
10(a) shows the morphology of CAPchitosan blends.
The morphology shows a highly deformed matrix as
CAP and chitosan form partially miscible blends. It
also includes a large number of elongated voids indicating resistance to removal of chitosan from the matrix. Figure 10(b) shows the morphology for blends
containing 10% NC. The surface shows a large number
of elongated voids caused by the debonded particles
spread through the entire surface area. The elongated
voids indicate deformation of the matrix and hence
the resistance for removal of particles. Thus, the modified NC has dispersed uniformly in the entire surface.
Figure 11(a)(e) shows the tensile fracture morphology of CAPchitosan blends using SEM. Figure
11(a) shows the fractured SEM micrograph of neat
chitosan. The micrograph shows a dense homogeneous structure with brittle characteristics, while that
for CAP [Figure 11(b)] shows a sheared matrix with
elongated voids due to tearing which indicates
higher ductility when compared with chitosan. The
blend without NC [Figure 11(c)] shows brittle failure
characterized by sheared matrix accompanied by
cavities left by debonded particles indicating the
interactions between CAP and chitosan. Similar
observations for chitosan blends with cellulose
ethers were reported by Yin et al.8
The SEM micrograph of the tensile fractured surface of the blend containing 6% NC is shown in [Figure 11(d)]. The micrograph shows a dense homogeneous interlocked surface with a large number of
elongated voids left by debonded particles. The
numbers of voids were found to increase at higher
filler content as shown in Figure 11(e).
E25
TABLE I
Variation of Water Uptake of Neat CAP, Chitosan, and
CAPChitosanNC blends
Sample NC (%)
CAP
Chitosan
NC0
NC2
NC3
NC4
NC6
NC8
NC10
Water absorption (%)

68.07
70.07
22.21
17.70
14.28
20.27
23.52
19.49
22.86
Water uptake
Table I shows the values of water absorption characteristics of CAPchitosan blends. chitosan and CAP
have similar water uptake characteristics (i.e., 70.07
and 68.07%, respectively). The blends (with no NC)
have a water uptake of 22.21%. This may be due to
the interactions between CAP and chitosan, which
form a network. The addition of NC up to 3% further reduces water absorbance as clay acts as a mechanical barrier. Addition of NC has further
improved the water barrier properties owing to the
tortuous path taken by the fluid and water absorption further drops down to 14.28%. However, at
higher content of NC (>4%), the water absorbency
increases. This may be due to the interaction
between the excess amine groups on the NC surface
with CAP and chitosan leading to a polymeric network. A similar observation was made by Zhang
et al.,38 in which case the surface groups of the claylike material interacted with modified chitosan leading to an increase in water absorbency.
Biodegradation studies
Figure 12 shows the plot of percentage weight loss
versus number of days for the CAPchitosan blends.
Chitosan is more biodegradable than CAP as the
hydroxyl groups in the latter are replaced by ester
groups. The blends show a retarded degradation for
the first 3 days but, thereafter, the biodegradation is
higher than either CAP or chitosan as observed in
the first 30 days. The addition of NC further lowers
the biodegradation up to 4% NC loading. This may
be due to the interaction between CAP and chitosan
with the amine groups of modified NC, which
restricts the segmental motion at the interface causing the effective path length and diffusion time to
increase. A similar observation has been made by
Rindusit et al.39 for methylcellulosemontmorillonite
(MMT) composites.
However, beyond 4% NC, the blend exhibits a
higher degradation than for lower NC loadings. The
addition of increased modified NC induced large
E26
GAURAV ET AL.
References
Figure 12 Biodegradation: variation of percentage weight

loss with number of days for CAPchitosanNC blends.
amorphous regions and these regions are easily accessible during the degradation process. A similar
observation was made by Wu and Wu,40 when the
biodegradation rates increased with 6% MMT when
compared with 3% MMT loading.
Further, there is an interrelation between water
uptake and degradability as higher water uptake
accelerates the degradation process. Thus, increase
in hydrophilicity increase leads to an increase in biodegradability. Thus, for water uptake, the blends
loaded with lower content of NC show a lower
uptake while blends loaded with >4% NC show
increased water absorption characteristics and hence
higher biodegradability.
CONCLUSIONS
CAP has been blended with chitosan along with
modified NC as reinforcing filler. The mechanical and
thermal properties were examined for NC variation.
The tensile strength reached an optimal value with
6% NC. The tensile modulus reduces as NC loading
increases, while the elongation at break increases.
Theoretical models used to analyze the obtained experimental values indicated interactions between the
blend components. Compressive strength improved
by 22% by the addition of NC, while the flexural
properties were unaffected for the nanocomposites.
Addition of NC enhanced the thermal stability as
indicated by an increase in char content. XRD studies
revealed exfoliation of NC in the blend. Water uptake
reduced to 14.28% on adding NC to the blends.
The authors thank the Department of Science and Technology (DST) for the financial assistance for carrying out this
work under the Green Chemistry Programme (20072010).
1. Angles, M. N.; Dufresne A. Macromolecules 2008, 33, 8344.

2. Li, Q.; Zhou, J.; Zhang, L. J Polym Sci B Polym Phys 2009, 47,
1069.
3. El-Tahlawy, K.; Hudson, S. M.; Hebeish, A. A. J Appl Polym
Sci 2007, 105, 2801.
4. El-Tahlawy, K.; Abdelhaleem, E.; Hudson, S. M.; Hebeish, A.
J Appl Polym Sci 2007, 104, 727.
5. Hasegawa, M.; Isogai, A.; Onabe, F.; Usuda, M.; Atalla, R. H.
J Appl Polym Sci 1995, 45, 1873.
6. Cai, Z.; Kim, J. J Appl Polym Sci 2009, 114, 280.
7. Altinisik, A.; Seki, Y.; Yurdakoc, K. Polym Compos 2009, 30,
1035.
8. Yin, J.; Lao, K.; Chen, X.; Khutoryanspiy, V. V. Carbohydr
Polym 2006, 63, 238.
9. Liu, C.; Bai, R. J Membr Sci 2005, 267, 68.
10. Shih, C.; Shieh, Y.; Twn, Y. Carbohydr Polym 2009, 79, 169.
11. Wibowo, A. C.; Misra, M.; Park, H. M.; Drzal, R. S.; Mohanty,
A. K. Compos A Appl Sci Manuf 2006, 37, 1428.
12. Huang, J.; Zhang, L.; Wang, X. J Appl Polym Sci 2003, 89, 1685.
13. Goswami, T. H.; Maiti, M. M. Polym Degrad Stabil 1998, 61,
335.
14. Shanmugharaj, A. M.; Rhee, K. Y.; Ryu, S. H. J Colloid Interface Sci 2006, 298, 854.
15. Menjoge, A. R.; Kulkarni, M. G. Int J Pharm 2007, 343, 106.
16. Balakrishnan, S.; Start, P. R.; Raghavan, D.; Hudson, S. D.
Polymer 2005, 46, 11255.
17. Sue, H. J.; Gam, K. T.; Bestaoul, N.; Suprr, N.; Clearfield, A.
Chem Mater 2004, 16, 242.
18. Sailaja, R. R. N.; Chanda, M. J. Appl Polym Sci 2001, 80, 863.
19. Willett, J. L. J Appl Polym Sci 1994, 54, 1685.
20. Bliznakov, E. D.; White, C. C.; Shaw, M. T. J Appl Polym Sci
2000, 77, 3220.
21. Hsieh, C. L.; Tuan, W. H. Mater Sci Eng 2005, 396, 202.
22. Zou, Y.; Wang, L.; Zhang, H.; Qian, Z.; Mou, L.; Wang, J.; Liu,
X. Polym Degrad Stabil 2007, 83, 87.
23. Isabella, F.; Micheline, B.; Alain, M. Polymer 1998, 39, 4773.
24. Siqueira, G.; Bras, J.; Dufresne, A. Polymer 2010, 2, 728.
25. Shi, X.; Hudson, J. L.; Spicer, P. P.; Tour, J. M.; Krishnamoorti,
R.; Mikos, G. Biomacromolecules 2006, 7, 2237.
26. Sagar, A. D.; Merrill, E. W. J Appl Polym Sci 1995, 58, 1647.
27. Sorrentino, A.; Gorasi, G.; Vittoria, V. Trends Food Sci Technol
2007, 18, 84.
28. Sithique, M. A.; Alagar, M.; Ali Khan. F. L.; Nazeer, K. P.
Malays Polym J 2011, 6, 1.
29. Tserki, V.; Matzinos, P.; Koppon, S.; Panayiotou, C. Compos A
2005, 36, 965.
30. Tserki, V.; Matzinos, M.; Panayiotou, C. Compos A 2006, 37,
1231.
31. El-Hefian, E. A.; Naset, M. M.; Yahaya, A. H. J Chem 2010, 7,
1212.
32. He, L.; Xue, R.; Yang, D.; Liu, Y.; Song, R. Chin J Polym Sci
2009, 27, 501.
33. Duan, W.; Chen, C.; Jiang, L.; Li, G. H. Carbohydr Polym
2008, 73, 582.
34. Rao, V.; Ashokan, P. V.; Shridhar, M. H. J Appl Polym Sci
2000, 76, 859.
35. Rao, V.; Ashokan, P. V.; Amar, J. V. J Appl Polym Sci 2002,
86, 1702.
36. Dong, Y.; Rwan, Y.; Wang, H.; Zhao, Y.; Bi, D. J Appl Polym
Sci 2004, 93, 1553.
37. Wang, L.; Wang, A. J Hazard Mater 2007, 147, 979.
38. Zhang, J.; Wang, Q.; Wang, A. Carbohydr Polym 2007, 68,
367.
39. Rindusit, S.; Jingjid, S.; Damrongsappul, S.; Tiptikaporn, S.;
Tapeichi, T. Carbohydr Polym 2008, 72, 444.
40. Wu, T.; Wu, C. Polym Degrad Stabil 2006, 91, 2198.

Publications Aashish Gaurav

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Publications Aashish Gaurav

Uploaded by

Copyright:

Available Formats

Article

Transesterication of Triglyceride to Fatty Acid Alkyl Esters

attractive as a renewable and environmentally friendly fuel option

Figure 1. Transesterication (methanolysis) of triglyceride molecule to

and FAAEs are then separated and puried. Rapeseed and

Received: September 4, 2013

dx.doi.org/10.1021/ef401790n | Energy Fuels 2013, 27, 68476857

Energy & Fuels

production process, especially in terms of reduction of water

equilibrium-limited. Other potential advantages of CD include

the literature4 and also conrmed via Aspen calculations in

Figure 2. Conguration A: traditional reactor separation ow sheet for biodiesel production.

dx.doi.org/10.1021/ef401790n | Energy Fuels 2013, 27, 68476857

Energy & Fuels

Figure 3. Conguration B: CD ow sheet.

Figure 4. Conventional ow sheet for biodiesel production from biomass feedstock.

dx.doi.org/10.1021/ef401790n | Energy Fuels 2013, 27, 68476857

Energy & Fuels

transesterication reaction followed by alcohol recovery in a distillation

length, diameter, and number of tubes, we also kept a watch on the

Figure 5. Transesterication of the triolein molecule to methyl oleate

Table 2. Arrhenius Parameters for the Transesterication

Because results in ref 11 indicated higher biodiesel yields at higher

dx.doi.org/10.1021/ef401790n | Energy Fuels 2013, 27, 68476857

Energy & Fuels

Figure 6. Flow sheet units: reactor separation ow sheet (conguration A).

Table 3. Reactor Conguration

3. PROCESS COMPARISONS (COST AND ENERGY)

dx.doi.org/10.1021/ef401790n | Energy Fuels 2013, 27, 68476857

Stream names are specied in Figure 2.

Table 4. Flow Sheet for the Reactor Separation: Conguration Aa

Energy & Fuels

dx.doi.org/10.1021/ef401790n | Energy Fuels 2013, 27, 68476857

Energy & Fuels

product stream price for technical-grade glycerol was set as

Figure 7. Flow sheet units: CD ow sheet (conguration B).

economic analyzer tool. The feed stream prices were set as

Stream names are specied in Figure 3.

optimal operation condition

heat duty (kW)

temperature (160 C),

mole RR (0.5), pressure

mole RR (0.5), pressure

utility cost ($/year)

dx.doi.org/10.1021/ef401790n | Energy Fuels 2013, 27, 68476857

Energy & Fuels

optimal operation condition

volume (1.80 m3)

temperature (150 C), pressure

stages (7), feed (2, 6)

mole RR (0.6), pressure (3 atm), per

utility cost ($/year)

equipment heating cost

equipment cooling cost

equipment total utility cost

considerably benet the cost of the overall process. Other merits

Because glycerol is a valuable byproduct of the biodiesel

dx.doi.org/10.1021/ef401790n | Energy Fuels 2013, 27, 68476857

Energy & Fuels

Figure 9. Temperature prole (liquid phase) in the CD column

requirements and associated costs into the total annual operating

equipment heating cost

equipment cooling cost

equipment total utility cost