Applied Energy 185 (2017) 985–997

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Use of reactive distillation in biodiesel production: A simulation-based

comparison of energy requirements and profitability indicators
Tuhin Poddar, Anoop Jagannath, Ali Almansoori ⇑
Department of Chemical Engineering, The Petroleum Institute, Abu Dhabi, P.O. Box 2533, United Arab Emirates

h i g h l i g h t s

 Simulation of two reactive distillation processes for biodiesel production is showcased.

 Column duty optimization and heat integration is performed for both processes.
 Economic analysis was performed using different profitability indicators.
 Heterogeneous-catalyzed process is more advantageous over alkali-catalyzed process.

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

Article history: The advent of biodiesel, as a viable alternative to replace crude-based diesel as transport fuel, has
Received 9 August 2015 prompted growing interest worldwide due to the need for low-emission fuels. In this work, two reactive
Received in revised form 8 December 2015 distillation processes using soybean oil as main feedstock along with the corresponding downstream
Accepted 14 December 2015
separation units are simulated: the first process involves a homogeneous alkali catalyst; whereas the
Available online 9 January 2016
second involves a heterogeneous catalyst. Both processes yield a high purity biodiesel product. In the
present work, ASPEN Plus v8.4 is used as the process simulation tool. The energy requirements of both
Keywords:
processes were evaluated based on the optimization of the distillation column duties and performing
Biodiesel
Reactive distillation
heat integration on the process streams. The optimization of the column duties was performed by
Process simulation analyzing the Column Grand Composite Curves (CGCC). The process streams were heat integrated, and
Column grand composite curves a Heat Exchanger Network (HEN) was designed to minimize utility consumption. Both processes were
Profitability analysis compared using profitability indicators such as Return-On-Investment (ROI), payback period and unit
production cost. The results show that the heterogeneous-catalyzed process is more profitable than
the alkali-catalyzed process for biodiesel production. The ROI, payback period and unit production cost
were 486%, 0.2 years and $0.712 per kg of biodiesel respectively. For the analysis, an annual production
capacity of 35.4 kilotonnes/year of biodiesel production was assumed.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction from triacylglycerols (mainly present in vegetable oil and fat)

In recent years, biofuels have emerged as a promising renew-
able energy source. Among the many biofuels, biodiesel has shown
great potential as a viable and popular alternative to standard
crude-based diesel fuel. There are numerous advantages for the
use of biodiesel: (1) synthesis from domestic renewable sources
(e.g., vegetable oil), (2) net carbon dioxide emissions reduction of
78% (on a lifecycle basis) compared to crude-based diesel, (3) envi-
ronmentally friendly given its biodegradable and non-toxic nature,
and (4) significantly improves engine exhaust emissions [1]. Bio-
diesel or mono-alkyl ester of long chain fatty acids can be obtained
⇑ Corresponding author. Tel.: +971 2 607 5544; fax: +971 2 607 5528.
E-mail address: aalmansoori@pi.ac.ae (A. Almansoori).
either by transesterification or esterification processes [2,3]. In
the former, the transesterification reaction occurs between triacyl-
glycerols and a low molecular weight alcohol (e.g., methanol) to
produce a complex mixture of Fatty Acid Methyl Esters (FAME
which is essentially biodiesel) and glycerol. In the esterification
or hydroesterification process, triacylglycerols are hydrolyzed to
free fatty acids, and then esterified with an alcohol (e.g., methanol)
to produce biodiesel and water.
The transesterification reaction requires a catalyst that can
either be in the form of a homogeneous alkali, a homogeneous acid,
or a heterogeneous alkali. In cases where the reaction is carried out
in supercritical conditions, there is no need for a catalyst [4]. For
supercritical conditions, the transesterification takes place using
alcohol at supercritical state (at a temperature higher than its

http://dx.doi.org/10.1016/j.apenergy.2015.12.054
0306-2619/Ó 2015 Elsevier Ltd. All rights reserved.
986 T. Poddar et al. / Applied Energy 185 (2017) 985–997
critical temperature). This eliminates the need for a catalyst. For
low molecular weight alcohol such as methanol, high temperatures
(up to 350 °C) and large alcohol-to-oil ratio (42:1) are required for
the reaction to achieve high conversion levels [4]. Many process
simulation approaches have been used to model biodiesel produc-
tion through transesterification reaction. These approaches have
included the use of different types of catalysts (i.e., alkali, acid or
heterogeneous alkali), feedstock (pure vegetable oil or waste cook-
ing oil), and reaction conditions (normal or supercritical) [1,4–13].
A standard simulation study comprises the simulation of the trans-
esterification reactor, followed by the downstream product purifi-
cation steps. One of the major developments in the simulation of
biodiesel processes consists of modeling the transesterification
reaction using a reactive distillation column. This is because the
transesterification between the triacylglycerols and alcohol is an
equilibrium limited reaction. Thus, the excess alcohol is added to
favor the reaction conversion towards biodiesel. This, in turn,
requires an expensive set-up to separate the large quantities of
unreacted alcohol from the biodiesel product. Integrating the
transesterification process and separating the reaction products
within the same unit, such as reactive distillation, overcomes the
limitations of equilibrium-based reactions. This allows for higher
biodiesel yields, by shifting the chemical equilibrium to the pro-
duct, compared to the conventional processes (reactor–separator
set-up). Apart from this, other advantages of reactive distillation
include reductions in the number of process units and savings in
the overall costs (no longer need of a reactor). Since the esterifica-
tion reaction is also equilibrium-based, reactive distillation is used
to simulate the esterification process of biodiesel production as
well.
Several studies have focused on the use of reactive distillation
to model the esterification reaction for biodiesel production
[2,3,14–18]. In all of these studies, it was found that sufficiently
high reaction conversion and product purity were obtained by
using reactive distillation. The esterification reaction for biodiesel
production primarily involves the usage of heterogeneous acid cat-
alysts. For the simulation of biodiesel production using reactive
distillation enabling transesterification reaction, most of the previ-
ous works include homogeneous alkali catalyst [19–21]; whereas
very limited work is available on heterogeneous catalyst [22].
Mueanmas et al. [19] studied the feasibility of using reactive distil-
lation for the production of biodiesel enabling transesterification
reaction using Aspen Plus simulation software. They found opti-
mum conditions for the design and operation of reactive distilla-
tion column in terms of molar ratio of alcohol to oil, reboiler
temperature, residence time, alcohol to feed location and reflux
ratio. Prasertsit et al. [20] studied a simple laboratory scale reactive
distillation set up of a packed column for the transesterification of
palm oil with methanol. They experimentally determined the opti-
mal operating conditions for the reactive distillation operation.
Their results were closely similar to the one obtained by Muean-
mas et al. [19]. Simasatitkul et al. [21] simulated and analyzed
the optimal conditions required for the design of reactive distilla-
tion column (in terms of molar ratio of alcohol to oil, reboiler tem-
perature, alcohol to feed location and reflux ratio) enabling
transesterification reaction between soybean oil and methanol
for the production of biodiesel using alkali catalyst. Gaurav et al.
[22] simulated the transesterification of triacylglycerols to FAME
by using a heterogeneous catalyst. They investigated and com-
pared the utility requirements and capital costs between the cat-
alytic reactive distillation and conventional (reactor–separator
set-up) process for biodiesel production. They concluded that the
catalytic reactive distillation process is more competitive alterna-
tive for biodiesel production in terms of capital and utility costs.
They also concluded that the cost of oil feedstock is the most sig-
nificant aspect in the total operating costs of both the biodiesel
production processes. Boon-anuwat et al. [23] designed and simu-
lated four different processes for biodiesel production using trans-
esterification reaction between soybean oil and methanol in Aspen
Plus simulation package. Four different processes simulated were
based on homogeneous alkali catalyst and heterogeneous acid cat-
alyst for both conventional (reactor–separator set-up) and reactive
distillation processes. They concluded that the reactive distillation
process using heterogeneous catalyst was the best among the con-
sidered processes as it had reduced energy consumption and
reduced number of units in its process design. Similar to the ester-
ification process, it is known from the Refs. [19–23], that the reac-
tive distillation based biodiesel production processes is more
superior and advantageous compared to the conventional biodiesel
processes. From the review of the literatures [19–23] for the simu-
lation of reactive distillation based biodiesel production processes
using transesterification reaction, we have noticed certain research
gaps in the current trend (reactive distillation based biodiesel pro-
duction processes) which need to be addressed. These gaps are
identified as follows.
1. Most of the studies on the reactive distillation process for bio-
diesel production have focused on the esterification reaction
for biodiesel production [2,3,14–18]. Despite the transesterifi-
cation process being the most important biodiesel production
process, only limited studies [19–23] exists for the use of reac-
tive distillation enabling transesterification reaction in the bio-
diesel production process.
2. Among the reactive distillation enabled transesterification reac-
tion biodiesel production processes, the homogeneous alkali
catalyst and heterogeneous catalyst are the well-known pro-
cesses. To the best of the authors’ knowledge, except the work
of Boon-anuwat et al. [23], a comprehensive comparison
(energy and economic based) between both processes (i.e.,
homogeneous alkali and heterogeneous catalyst), is currently
lacking in the literature. Boon-anuwat et al. [23] provides a
comparison between biodiesel production using homogeneous
alkali catalyst and heterogeneous catalyst. However, their work
does not include cost comparisons between both processes.
3. Although the previous literature work [19–23] does compute
the energy requirements and finds out the optimal conditions
necessary for the reactive distillation based biodiesel process
using transesterification reaction (using homogeneous alkali
catalyst or heterogeneous catalyst), there is no work in the lit-
erature which systematically carries out optimization
approaches to minimize the energy requirements for the reac-
tive distillation based biodiesel process enabling transesterifi-
cation reaction. The energy reduction approaches for the
downstream biodiesel product separation steps have also been
overlooked by the previous literature work, as most of it con-
centrates primarily on the energy aspects and optimum condi-
tions for the main reactive distillation column (reactive
distillation based biodiesel process enabling transesterification
reaction).
Addressing all of the above mentioned issues/gaps constitutes
the motivation of this present work. Therefore, the objective of this
work is to find the more suitable reactive distillation process for
biodiesel production using a process simulation approach. The
transesterification reaction pathway has been chosen for the bio-
diesel production. In this study, AspenPlusTM by AspenTech has
been used as the simulation software. Both (homogeneous alkali
and heterogeneous catalyst) processes are compared in terms of
energy efficiency, cost effectiveness and profitability to determine
the most suitable one. The following assumptions were considered
for the simulation purposes: (1) the acceptable final product
biodiesel purity must be at least 98% (molar) pure while the
 T. Poddar et al. / Applied Energy 185 (2017) 985–997
by-product glycerol purity must be at least 92% pure, (2) The bio-
diesel plant annual capacity is 35.4 kilotonnes per year of biodiesel
and operates 8760 h per year, (3) for the purpose of economic eval-
uation, the feedstock for used is pure soybean oil with very low
free fatty acids content (less than 0.3%), thus avoiding the need
for oil pretreatment (4) for the purpose of simulation, triolein
(C57H104O6) is used to represent the triglyceride in the soybean
oil. The usual composition (wt.%) of soybean oil is approximately
23.7% oleic acid, 55.5% linoleic acid, and 4.9% linolenic acid [24].
Since these three major unsaturated fatty acids contain 18 carbon
atoms each, soybean oil can be reasonably represented by a simple
triglyceride namely triolein [22]. Similarly, methyloleate
(C19H36O2) is used to represent FAME or biodiesel (4) the grass-
root study [25] of biodiesel production was assumed. The article
is structured as follows. Section 2 describes the approach taken
to compare both the processes. Section 3 presents the descriptions
for both the base case processes. A brief description of the opti-
mization of the distillation columns and heat integration are given
in Sections 4 and 5 respectively. Section 6 covers the economic
analysis approach used to compare both processes. Section 7 pro-
vides a detailed discussion and comparison between both pro-
cesses. Lastly, the major conclusions of this research are
highlighted in Section 8.
2. Methodology
The approach taken to assess and compare the two biodiesel
production processes in this work are explained in this section.
The first comparison step between both processes is based on
energy requirements. Accordingly, two approaches have been
considered, namely the distillation column optimization and heat
integration. Distillation is one of the most energy intensive and
least energy efficient separation processes used in industry [26].
Biodiesel production is characterized by the presence of distillation
columns; which contribute significantly to the capital cost and
dominate the energy requirements. Any step towards the optimiza-
tion of the distillation columns duties results in energy reduction,
decrease in utility requirement, increase in energy efficiency, and
reduced capital cost. In this work, for both processes, the distillation
column analysis and duty optimization were done using thermal
analysis (Column Grand Composite Curve (CGCC)) [27–30].
Through this analysis, the profiles generated by the column target-
ing tool (available in the Aspen Plus) were studied. Additionally, a
heat integration analysis was performed in both processes by set-
ting up a Heat Exchanger Network (HEN). This was done to: (1)
maximize energy recovery among the process streams and col-
umns, (2) lower the dependency on external utilities, and (3)
improve the energy efficiency of the overall process. The Aspen
Energy Analyzer software, included in the AspenTech package,
was used for the heat integration analysis. The second comparison
step between both processes consists of the process economics and
profitability indicators. The Activated Analysis for Economics tool, a
built-in tool in Aspen Plus package, was used to obtain the capital
and operational costs involved in biodiesel production. These pro-
cess economics were then used to calculate profitability indicators;
which formed the basis for the economic comparison.
The originality and contribution of this work are highlighted as
follows
1. The present work compares two biodiesel production processes
(using reactive distillation) in a comprehensive manner by con-
sidering both energy and costs comparisons and determines the
most profitable between the two. Unlike most other work, the
present work compares the energy and cost aspects of the
entire process flowsheet including the biodiesel transesterifica-
tion and downstream product and by-product separation steps.
987
2. To reduce the energy requirements in the biodiesel production
process, the distillation column duties, for each process, were
optimized using the CGCC analysis. This was followed by con-
sidering heat integration among the process streams and col-
umns, and an HEN was set up to minimize overall energy
consumption.
Such energy reduction measures followed by a profitability
analysis have been scarcely considered in the literature. From the
results (exhibited later) of column optimization and heat integra-
tion, significant energy reductions in both processes were
observed. Most process flowsheets for biodiesel production usually
do not focus much on the energy optimization and heat integration
aspects. Thus, the present work is unique, in that, it considers col-
umn optimization, heat integration and economic analysis to iden-
tify the best process for biodiesel production using reactive
distillation enabling transesterification reaction pathway.
3. Process description of base cases
3.1. Alkali-catalyzed homogeneous process
Fig. 1 shows the base case flow diagram for the homogeneous
alkali-catalyzed process for biodiesel production using reactive
distillation. Two feed streams, one of oil feedstock (soybean oil rep-
resented by triolein (simple triacylglycerol)) and another of metha-
nol at room temperature and atmospheric pressure were heated to
60 °C before entering the reactive distillation column RD (see
Fig. 1). The methanol and oil feed were fed in a molar ratio of
6:1 into the RD column. The RD column is fitted with 10 stages,
and specified with a molar reflux ratio of 0.2 and a boil-up ratio
of 0.325. Both feed streams enter the column at the 3rd stage,
and the column works at atmospheric pressure. For the alkali-
catalyzed process simulation, the fluid package was specified to
be UNIFAC. The reaction pathway and kinetic parameters for the
transesterification reaction were taken from Samasatitkul et al.
[21]. A standard RADFRAC column (available in Aspen Plus) was
used to simulate the RD column. Normally sodium hydroxide is
used as a catalyst for this process, and it is usually fed into the
RD unit along with the alcohol at a flowrate specified as 1 wt.%
of the oil feedstock flowrate. In this work, however, the sodium
hydroxide catalyst was not explicitly modeled in the simulation
(Fig. 1). This was because the reaction was setup in the simulation
using the kinetic data [21], which took into account the behavior of
the sodium hydroxide catalyst in the transesterification reaction to
form biodiesel and glycerol. In the RD column, the transesterifica-
tion reaction took place to form methyloleate (biodiesel) and glyc-
erol. Close to 100% of the unreacted methanol was collected at the
top of the column as a distillate (DISTOP stream as shown in Fig. 1)
and recycled to the alcohol stream. The DISBOT stream, which
comprised of the reaction products (methyloleate and glycerol)
and some unreacted methanol, formed the bottom product. The
main unit operation design specifications have been summarized
in Table 1.
The DISBOT stream is then pumped to 4.93 atm before entering
a liquid–liquid extractor (LLE column as shown in Fig. 1). The pur-
pose of this liquid–liquid extraction was to separate methyloleate
(biodiesel) from glycerol and methanol. The solvent used for
extraction was hexane; which entered the LLE column at a flow
rate of 10 kmol/h, temperature of 25 °C, and pressure of 0.99 atm.
The extractor is outfitted with 8 stages. The methyloleate, hexane
and very minute quantities of unreacted methanol and glycerol
were collected at the top of the extraction column as FAME1
stream (see Fig. 1), while the glycerol and some unreacted metha-
nol were collected at the bottom of the extractor at GLY1 stream
988 T. Poddar et al. / Applied Energy 185 (2017) 985–997

Fig. 1. Process flowsheet for base case homogeneous alkali-catalyzed process.

Table 1 distillate mixture of methanol and hexane was obtained at the top.
Summary of specifications for the main unit operations for the alkali-catalyzed The final methyloleate (biodiesel) product was brought to 25 °C
process and heterogeneous process.
and 1 atm. Table A.1 (see Appendix) gives the details of the key
Alkali-catalyzed Heterogeneous-catalyzed process streams for process shown in Fig. 1 (homogeneous alkali-
process process catalyzed process). The design specifications for the unit opera-
Reactive distillation tions shown in Fig. 1 (homogeneous alkali-catalyzed process) are
Reflux ratio 0.2 1.5 summarized in Table 1.
Number of stages 10 7
Boilup ratio/distillate rate 0.325 13.61
(kmol/h) 3.2. Heterogeneous-catalyzed process
Pressure (atm) 1 3
Liquid–liquid extraction The thermodynamic fluid package used for the heterogeneous-
Number of stages 8 N/A
catalyzed process was UNIFAC, while the kinetic parameters for the
Biodiesel recovery N/A transesterification reaction were obtained from Gaurav et al. [22].
Reflux ratio 2
The solid base catalyst used in this process was calcium oxide sup-
Number of stages 10
Distillate to feed ratio 0.45 ported on alumina (20 wt.% CaO (calcium oxide) and 80 wt.% Al2O3
Pressure (atm) 0.02/0.09 (alumina)) [22]. Fig. 2 shows the base case flow diagram for the
Glycerol recovery N/A heterogeneous catalyzed biodiesel production process using reac-
Reflux ratio 2 tive distillation. Similar to the alkali catalyzed process, the catalyst
Number of stages 10 was not explicitly incorporated into the simulation as it was repre-
Bottoms to feed ratio 0.8 sented in the transesterification reaction through the kinetic data
Pressure (atm) 0.15/0.20
[22]. Two feed streams for the oil feedstock (triolein) and methanol
Decanter N/A (at 25 °C and 1 atm) were pumped to 3.2 and 3.7 atm respectively
Temperature (°C) 25
before entering the reactive distillation column RD2 (see Fig. 2). In
Pressure (atm) 1
RD2, transesterification reaction occurs and biodiesel and glycerol
were produced. The RD2 column is fitted with 7 stages, and had a
reflux ratio and distillate rate of 1.5 and 13.61 kmol/h respectively.
(see Fig. 1). The GLY1 stream would ideally include the sodium
The unreacted methanol collected at the top as a distillate stream
hydroxide catalyst. This stream would then be fed to a neutraliza-
DISTP (see Fig. 2) and recycled to the alcohol feed stream. The bot-
tion reactor to remove sodium hydroxide. This is done by reacting
toms stream DISBT (see Fig. 2) leaves the column at high temper-
GLY1 stream with phosphoric acid. The phosphoric acid reacts the
ature (256.2 °C) and pressure (3 atm). Subsequently, DISBT was
sodium hydroxide, present in GLY1, as sodium phosphate; which is
sent through an expander and a cooler to bring the temperature
later removed using a gravity separation process. The neutraliza-
back down to 25 °C and the pressure to 1 atm. The final separation
tion reactor and gravity separation process were not modeled in
of the products was facilitated by gravity separation (decanter) due
the present simulation (Fig. 1) as the focus was more on the reac-
to the difference in the densities of biodiesel and glycerol. The
tion products namely methyloleate (biodiesel) and glycerol. How-
decanter was set to operate at 25 °C and 1 atm. The methyloleate
ever, for a more comprehensive and realistic simulation of process
(biodiesel) was collected as the first liquid (stream FAME in
flowsheet for biodiesel production using homogeneous alkali cata-
Fig. 2) at a molar purity >98%; while the glycerol was collected
lyst, the neutralization reactor and gravity separation process
as second liquid (stream GLY in Fig. 2) with a molar purity >92%.
could be simulated along with the other separation units.
Both streams exit the decanter at room temperature and pressure.
FAME1 and GLY1 streams were subjected to further distillation
Table A.2 (see Appendix) gives the key process streams for process
processes to obtain pure biodiesel and glycerol. The first distilla-
shown in Fig. 2 (heterogeneous catalyzed process). The design
tion column, labeled DIST1 (Fig. 1), was used to separate the
specifications for the unit operations shown in Fig. 2 (heteroge-
methanol and glycerol. The GLY1 stream enters at stage 7 of the
neous catalyzed process) are summarized in Table 1.
DIST1 column. This column has 10 stages, worked at a reflux ratio
of 2, and had bottoms-to-feed ratio of 0.8. The FAME1 stream
entered column DIST2. This column also had 10 stages and worked 4. Optimization of distillation columns
at a reflux ratio of 2. However, the feed FAME1 enters the column
on the 5th stage. Also, DIST2 had a distillate to feed ratio of 0.45. As mentioned in Section 2, the optimization of distillation
Emerging from the column DIST1 were the distillate stream with columns is done through thermal analysis. The thermal analysis
100% methanol purity and a glycerol rich bottoms stream (molar of the column is useful to identify energy targets (condenser
purity >92%). From DIST2, methyloleate (biodiesel) with high pur- and reboiler heat load) and column design modifications required
ity (molar purity >98%) was obtained as the bottom stream, and a to improve energy savings. Thermal analysis involves the
 T. Poddar et al. / Applied Energy 185 (2017) 985–997 989

Fig. 2. Process flowsheet for base case heterogeneous catalyzed process.

construction of the temperature–enthalpy curve (temperature-H 11

CGCC plot) or stage–enthalpy curve (stage-H CGCC plot) known 10
as the CGCC. The CGCC represents the thermal profile (also called 9
ideal profile) of an ideal column operating according to the rever-
8
sible column approach [31]. In the reversible column approach, the
7
Actual Profile
column operates at certain limiting conditions, also known as min-

Stage
imum thermodynamic conditions. The minimum thermodynamic 6 Ideal Profile
conditions involve imposing minimum reflux conditions at each 5
stage; such that the column has: (1) infinite number of stages 4
and (2) infinite number of side exchangers (heaters and coolers)
3
matching the temperature at each stage. The reversible column,
2
which is the most energy efficient column, can be used to identify
column design modifications to reduce energy consumption. In the 1
0 1000000 2000000 3000000 4000000 5000000 6000000
limiting conditions of minimum thermodynamic conditions,
Enthalpy Deficit (kJ/hr)
energy can be supplied to the column along the ‘ideal profile’
instead of column condenser and reboiler temperatures. In a nor- Fig. 3. Base case stage-H CGCC plot for RD column.
mal column operation, energy is supplied to the column at con-
denser and reboiler temperatures; which represents the ‘actual
profile’ in the CGCC. Thus, the CGCC shows the ideal and actual 11

enthalpies (ideal and actual profiles) at each stage of the column 10

along with the condenser and reboiler duties. However, minimum
9
thermodynamic conditions, seem reasonable only for binary com-
ponents having sharp separations. For columns with multiple com- 8

ponents and non-sharp separations, a condition known as Practical 7

Actual Profile
Stage

near Minimum Thermodynamic Condition (PNMTC) is applied. In 6

this approach, the enthalpy calculation for the ‘ideal profile’ takes Ideal Profile
5
into account the practical losses from a realistic column design
such as: pressure drops, heat and mass transfer losses [32]. Aspen 4
Plus performs thermal analysis to generate the CGCC for rigorous 3
column designs usually based on the PNMTC [29]. The CGCC anal-
2
ysis involves understanding the temperature–enthalpy curve or
stage–enthalpy curve of a column and to make the necessary 1
0 20000 40000 60000 80000 100000 120000 140000 160000
changes in the column design such that the column duties (con- Enthalpy Deficit (kJ/hr)
denser and reboiler) are reduced. The column design aspects usu-
ally manipulated (based on CGCC analysis) are: feed stage Fig. 4. Base case stage-H CGCC plot for DIST1 column.
location, reflux ratio, feed temperature, and addition of side con-
denser/reboiler. For more information regarding the analysis of enthalpy change occurs on the condenser side of the stage-H CGCC
CGCC curves, the reader can refer to the works in Refs. [27–30]. plot. This indicates that the column feed is introduced too high and
should be lowered. The horizontal distance between the ordinate
4.1. Alkali-catalyzed homogeneous process axis and stage-H CGCC pinch point (co-ordinates: 4,958,762 kJ/h,
3rd stage) shows that a reduction in reflux ratio is also possible
Table 1 shows the base case operating conditions for the three (see Fig. 3). Similarly, Fig. 4 illustrates that there is a gap between
distillation columns: (1) the reactive distillation column (RD), (2) the stage-H CGCC pinch point (co-ordinates: 87,253 kJ/h, 6th stage)
the glycerol separation column (DIST1) and (3) the methyloleate and the stage axis (ordinate axis). Also, the feed is introduced too
separation column (DIST2) in the homogeneous alkali catalyzed low into the column as a sharp enthalpy change occurs close to
process for biodiesel production. Figs 3–5 show the CGCC stage– the reboiler side. Hence the feed stage can be moved further up.
enthalpy deficit curves for the base case process generated by Fig. 5 shows that there is little scope for reflux ratio reduction as
ASPEN Plus for RD, DIST1 and DIST2 respectively. In Fig. 3, a sharp the CGCC pinch point is close to the vertical axis (although
990 T. Poddar et al. / Applied Energy 185 (2017) 985–997

11 6. Economic analysis approach

10
The first step in the economic analysis of both processes was to
9
calculate the Total Installation Cost (TIIC); which is the capital cost
8
of the equipments involved in biodiesel production. These were
7 directly obtained from the Activated Analysis tool in Aspen Plus.
Stage

6 Alternatively, the equipment installation costs can be obtained

5
4
3 Ideal Profile
2
1
0 500000 1000000 1500000 2000000 2500000
Enthalpy Deficit (kJ/hr)
Fig. 5. Base case stage-H CGCC plot for DIST2 column.
reduction of reflux ratio is still favorable). However, large area
above the CGCC pinch point (co-ordinates: 83,876 kJ/h, 4th stage)
(area between the ideal and actual enthalpy profiles) shows scope
for addition of a side reboiler/heater.
4.2. Heterogeneous-catalyzed process
Table 1 shows the base case operating conditions for the reac-
tive distillation column, RD2, for the heterogeneous catalyzed pro-
cess. This column includes 7 stages and the purity of methyloleate
in the FAME stream (see Fig. 2) meets the product specification
(>98 mol%). Fig. 6 shows the stage-H CGCC plot for the base case.
The gap between the stage-H CGCC pinch point (co-ordinates:
935,835 kJ/h, 4th stage) and the ordinate axis represents excess
heat. This is an indication that modifying the reflux ratio can
reduce the condenser and reboiler duties.
5. Heat integration
After the optimization of the columns, the next energy opti-
mization step is to carry out a heat integration analysis among pro-
cess streams and columns by setting up a HEN. Accordingly, the
entire process flowsheet was imported to Aspen Energy Analyzer
to retrieve the information regarding the hot and cold streams
for the HEN grid diagram. The Aspen Energy Analyzer default val-
ues were used for the inlet and outlet temperatures of the utilities.
The heat integration analysis for the process streams and columns
(in the flowsheet) was carried out in accordance with the concept
of pinch analysis.
8
7
using standard mathematical expressions available in the litera-
ture [33,34]. The next step was to calculate the Total Indirect Cost
Actual Profile
(TIC), which comprises several indirect capital costs such as: site
preparation, service facilities, allocation costs, engineering and
supervision. Next, the Direct Permanent Cost (DPC) was calculated
combining both the TIIC and TIC. The Fixed Capital Investment
3000000 3500000
(FCI) was estimated to be 20% of the DPC. The Total Capital Invest-
ment (TCI) is calculated as the sum of the FCI and the working cap-
ital (WC). The WC is assumed to be 15% of the FCI. TCI is one of the
key economic indicators used to compare the cost-effectiveness of
both processes [5,35].
The annual production cost is defined as the operational cost
incurred by the manufacturer in producing biodiesel. This is
mainly a function of raw material costs and utilities. The raw mate-
rial costs comprised the cost of soybean oil, methanol, catalyst and
other associated chemicals involved in the biodiesel production
process. The following feedstock unit prices were considered for
the analysis: for soybean oil, $0.6/kg [36], and $0.35/kg for metha-
nol [37]. Additionally, for the homogeneous alkali-catalyzed pro-
cess, the unit cost of sodium hydroxide catalyst is assumed to be
$0.35/kg, and $0.5/kg for the LLE solvent, hexane. The amount of
catalyst required for homogeneous alkali-catalyzed process was
estimated from the flowrate of oil feedstock. On the other hand,
for the heterogeneous catalyzed process the catalyst cost of
$100/kg was obtained from Gaurav et al. [22]. The cost coefficients
(on a $/kJ basis) for the utilities are: $2.125  10 7/kJ, $2.5  10 6/
kJ and $2.739  10 6/kJ for cooling water, high pressure steam, and
refrigerant, respectively. The annual income is the manufacturer’s
earnings from the product sales. In this paper, the income was cal-
culated based on biodiesel (product) and glycerol (by-product)
sales. The selling price of biodiesel and glycerol were specified to
be $0.79/kg [38], and $1.1/kg [11] respectively.
From the previous discussed costs (FCI, TCI, annual production
cost and annual product income), the profitability indicators such
as ROI, Payback period, and the unit production cost are calculated.
ROI represents the profit rate on an investment and is calculated as
the ratio of annual net profit to the TCI [33]. The annual net profit is
the difference between the annual income and annual production
cost. The payback period is the time required, after plant start-
up, to recover the FCI. The shorter the payback period, the more
profitable is the process [33]. This can be calculated as the ratio
of FCI to annual net profit. The unit production cost of biodiesel
is an important profitability criterion [22] and it is calculated from
the annual production cost and production rate.

6 7. Results and discussion

5 7.1. Optimization of distillation columns in homogeneous alkali-

Stage

Actual Profile catalyzed process

4
Ideal Profile
3 Based on the observations presented in Section 4.1, changes
were implemented on the RD, DIST1 and DIST2 columns and this
2 constituted the optimized case. For the RD column, the feed stage
was increased from 3 to 5. The reflux ratio was decreased to 0.01
1
0 500000 1000000 1500000 2000000 2500000 3000000 3500000 from 0.2. By incorporating these changes, the condenser duty
Enthalpy Deficit (kJ/hr)
was reduced from 478,677 kJ/h in the base case to 414,325 kJ/h
(13.4% reduction) in the optimized case. Similarly the reboiler duty
Fig. 6. Base case stage-H CGCC plot for RD2 column. was reduced from 627,253 kJ/h in the base case to 562,871 kJ/h
 T. Poddar et al. / Applied Energy 185 (2017) 985–997 991

(10.3% reduction) in the optimized case. In the DIST1 column, the 11

feed stage was lowered from 7 to 2 and the reflux ratio was 10
decreased from 2 to 0.2. As a result, the condenser duty changed
9
from 130,848 kJ/h in the base case to 42,773 kJ/h (67.3% reduc-
tion) in the optimized case. Likewise, the reboiler duty decreased 8
substantially from 139,041 kJ/h in the base case to 733 kJ/h 7 Optimized Case
(99.5% reduction) in the optimized case. For DIST2, however, a side

Stage
reboiler was added to the column at stage 5 with a duty of 6 Base Case
1,055,056 kJ/h and the reflux ratio was reduced to 0.5 from 2. This 5
reduced the condenser and the reboiler duty to 894,595 kJ/h and
4
969,187 kJ/h respectively. However, the temperature of the PRO-
DUCT stream (see Fig. 1) containing FAME product had a tempera- 3

ture of 255 °C; which was higher than the decomposition 2

temperature of FAME. To circumvent this issue, the side reboiler
1
was removed and instead feed cooling was considered. The feed 20000 0 20000 40000 60000 80000 100000 120000 140000 160000
to the column DIST2 was cooled to 79.4 °C from 101.6 °C. Addition- Enthalpy Deficit (kJ/hr)
ally, the optimization module in Aspen Plus was activated for the
DIST2 column. The main objective of the optimization was to min- Fig. 8. Comparison of the stage-H CGCC plot for base case and optimized case
imize both the column condenser and reboiler heat duties. The design for DIST1 column.
optimization was achieved by varying column specifications,
namely reflux ratio and distillate flow. A constraint was imposed
11
on the DIST2 column optimization such that the temperature of
the FAME in the bottoms stream would be no higher than 250 °C. 10

Another constraint was imposed on DIST2 to ensure that the purity 9

(mole fraction) of FAME in the PRODUCT stream (see Fig. 1) would
8
always be at least 98%. The implementation of these changes
7
reduced significantly the condenser duty from 1,789,018 kJ/h in Stage
the base case to 508,123 kJ/h (71.6% reduction) in the optimized 6
case. Correspondingly, the reboiler duty decreased from 5
2,918,666 kJ/h in the base case to 1,349,840 kJ/h (53.8% reduction) Optimized Case
4
in the optimized case. The optimal reflux ratio and the distillate-to- Base Case
feed ratio for column DIST2 were 0.25 and 0.444 respectively. 3
Figs. 7–9 show the comparison of the stage–enthalpy deficit curve 2
of the base case and optimized case design for the columns RD,
1
DIST1 and DIST2 (in the homogeneous alkali catalyzed process) 500000 0 500000 1000000 1500000 2000000 2500000 3000000 3500000
respectively. From Figs. 7–9, it can be observed that the stage– Enthalpy Deficit (kJ/hr)
enthalpy deficit profile for the optimized case is much closer to
the vertical axis than the base case design. This means that in Fig. 9. Comparison of the stage-H CGCC plot for base case and optimized case
the optimized case, the column requires lesser energy than the design for DIST2 column.

base case. Table 2 shows a comparison between the operating con-

ditions of the base and optimized cases for columns RD, DIST1 and
total exergy losses were reduced by 91.2%. This is from 29,540 kJ/
DIST2.
h in base case to 2,593 kJ/h in the optimized case. Fig. 12 also
Figs. 10–12 show the comparison of the stage–exergy loss pro-
shows that the optimized case operates with less exergy losses
file for the base and optimized cases for the RD, DIST1 and DIST2
than the base case in the feed stage and condenser. However, there
columns respectively. Fig.11 shows that the base case design
were more exergy losses in the reboiler for the optimized case
include large exergy losses (accessible work), especially at the feed
compared to the base case. Overall, the total exergy losses in the
stage and reboiler. In the optimized case for DIST1 column, the
optimized case for DIST2 were 220,011 kJ/h which was lesser than
1,015,392 kJ/h (by 78.3%) in the base case. For the RD column, the
11
total exergy losses in the optimized case were 583,110 kJ/h which
10 was only marginally better (by 1.5%) than the exergy losses of
9 592,035 kJ/h in the base case. This is evident from the similarity
Base Case of the stage–exergy profiles for the base and optimized case exhib-
8
Optimized Case ited in Fig. 10.
7
Stage

6
7.2. Optimization of distillation columns in heterogeneous-catalyzed
5
process
4

3 Upon closely observing the stage-H CGCC plot in Section 4.2, the
reflux ratio is reduced from 1.5 to 0.75. As the reflux ratio is
2
reduced, the condenser duty decreases from 1,158,020 kJ/h in
1 the base case to 816,650 kJ/h (29.5% reduction); whereas the
1000000 0 1000000 2000000 3000000 4000000 5000000 6000000
reboiler duty decreases from 3,072,670 kJ/h in the base case to
Enthalpy Deficit (kJ/hr)
2,709,470 kJ/h (11.8% reduction). Fig. 13 shows the stage-H CGCC
Fig. 7. Comparison of the stage-H CGCC plot for base case and optimized case plot for the case when the reflux ratio was decreased to 0.75. The
design for RD column. purity of the methyloleate in the FAME stream (see Fig. 2) is similar
992 T. Poddar et al. / Applied Energy 185 (2017) 985–997

Table 2
Comparison of operating conditions for base case and optimized case for RD, DIST1 and DIST2 columns.

Conditions RD column DIST1 column DIST2 column

Base case Optimized case Base case Optimized case Base case Optimized case
No. of stages 10 10 10 10 10 10
Feed stage 3/3 5/5 7 2 5 5
Feed temperature (°C) 60/60 60/60 132.72 132.72 112.29 80
Reflux ratio 0.2 0.01 2 0.2 2 0.25
Condenser duty (kJ/h) 478,677 414,325 130,848 42,777 1,789,018 508,123
Condenser temperature (°C) 64.54 64.54 22.83 22.86 19.5 19.79
Condenser pressure (atm) 1 1 0.15 0.15 0.02 0.02
Distillate rate (kmol/h) 11.30 11.30 1.15 0.8620 11.12 10.97
Reboiler duty (kJ/h) 627,253 562,871 139,041 733 2,918,666 1,349,840
Reboiler temperature (°C) 127.46 127.46 154.25 102.4 254.28 189.73
Reboiler pressure (atm) 1 1 0.2 0.2 0.09 0.09
Bottoms rate (kmol/h) 20.48 20.48 1.29 4.91 13.59 13.74

11 11

10 10
Optimized Case
9 9
8 Base Case Optimized Case
8
7 Base Case
7
Stage

Stage
6
5
5
4
4
3

2 3

1 2
100000 0 100000 200000 300000 400000 500000 600000
1
Exergy Loss (kJ/hr) 50000 0 50000 100000 150000 200000 250000 300000 350000 400000 450000 500000
Exergy Loss (kJ/hr)
Fig. 10. Comparison of the stage–exergy loss profile of base case and optimized
case design for RD column. Fig. 12. Comparison of the stage–exergy loss profile of base case and optimized
case design for DIST2 column.

11

10 8

9
7
8
6
7
Stage

6 5
Stage

Actual Profile
5 Optimized Case
4
Ideal Profile
4 Base Case

3 3

2
2
1
2000 0 2000 4000 6000 8000 10000 12000 14000 1
Exergy Loss (kJ/hr) 0 500000 1000000 1500000 2000000 2500000 3000000
)
Enthalpy Deficit (kJ/hr
Fig. 11. Comparison of the stage–exergy loss profile of base case and optimized
case design for DIST1 column. Fig. 13. Base case stage-H CGCC plot for RD2 column for reflux ratio = 0.75.

to that in the base case. Fig. 13 also shows sharp enthalpy changes
close to the reboiler side, which is an indicative of sub-cooled feed.
Hence, the feed needs to be pre-heated before being entering the
reactive distillation column RD2 to reduce the column duties.
The feed temperature (for both oil and methanol streams) was
increased to 75 °C. The reboiler duty further decreased to
2,083,390 kJ/h, while the condenser duty increased slightly to
819,905 kJ/h. However, the purity of the methyloleate in the
FAME stream reduced considerably (from 98.5 mol% to 97 mol%).
This required modification of the column specifications to ensure
that the methyloleate content in the FAME stream was maintained.
To ensure this in a systematic manner, the optimization module in
Aspen Plus was activated with the objective of minimizing the con-
denser and reboiler duties. The column specifications varied were
reflux ratio and distillate flowrate. The bounds on these specifica-
tions (reflux ratio and distillate flowrate) were obtained by sensi-
tivity analysis performed on the column duties by varying each
of these specifications. In addition, the number of column stages
was increased from 7 to 10. An optimization constraint was also
imposed such that the purity of methyloleate in the FAME stream
 T. Poddar et al. / Applied Energy 185 (2017) 985–997 993

11 would be at least 98 mol%. After implementing these changes, the

10 results showed a condenser and reboiler duty of 472,313 kJ/h and
Optimized Case 1,835,575 kJ/h respectively; and the purity of methyloleate in
9
Base Case FAME stream was 98.5 mol%. However, after optimization, the
8 reflux ratio increased to 2. This case was now considered as the
7 optimized case. Fig. 14 shows the comparison between the stage-
H CGCC plot of the base and optimized case design for the reactive
Stage

6
distillation column RD2 (in the heterogeneous catalyzed process).
5 Table 3 shows a comparison between the operating conditions of
4 the base and optimized case design. Additionally, Fig. 15 shows
3
the comparison between the stage–exergy loss curve of the base
and optimized case design. From Fig. 15, it can be observed that
2
the exergy losses in the optimized case design were less than that
1 of the base case for all stages; including condenser and reboiler. For
500000 0 500000 1000000 1500000 2000000 2500000 3000000 3500000
the RD2 column, the optimized case reduced the total exergy losses
Enthalpy Deficit (kJ/hr)
by about 48.3%, (from 1,313,638 kJ/h in base case to 679,596 kJ/h in
Fig. 14. Comparison of the stage-H CGCC plot for base case and optimized case the optimized case). Figs. 16 and 17 show the optimized case pro-
design for RD2 column. cess flowsheets for biodiesel production using alkali and heteroge-
neous catalyzed processes respectively.
Table 3
Comparison of operating conditions for base case and optimized case for RD2 column 7.3. Heat integration
for heterogeneous-catalyzed process.

Conditions Base case Optimized case Table A.3 (Table A.4) (see Appendix) and Fig. 18 (Fig. 19) depict
No. of stages 7 10 the stream information and HEN grid diagram for the optimized
Feed stage 2/6 2/6 case process flowsheet for the homogeneous alkali-catalyzed pro-
Feed temperature (°C) 25.28/63 75 cess (heterogeneous-catalyzed process) respectively. It should be
Reflux ratio 1.5 2
noted that the HEN diagrams (as shown in Figs. 18 and 19) were
Condenser duty (kJ/h) 1,097,590 472,313
Condenser temperature (°C) 95.4 95.4 only one of the possible designs, and that, multiple designs were
Condenser pressure (atm) 3 3 also possible for the same process. For the homogeneous alkali cat-
Distillate rate (kmol/h) 13.6078 4.5736 alyzed process (Fig. 18), the heating and cooling requirements
Reboiler duty (kJ/h) 3,072,670 1,835,580
were 467,048 kJ/h (32.11 kW h/tonne of biodiesel) and
Reboiler temperature (°C) 256.2 248.3
Reboiler pressure (atm) 3 3 1,090,165 kJ/h (74.94 kW h/tonne of biodiesel) respectively (7 heat
Bottoms rate (kmol/h) 18.18 18.18 exchangers, 4 utility coolers and 1 utility heater). The heating
requirements were satisfied using high pressure steam, whereas
11
the cooling requirements were satisfied partly (582,042 kJ/h) by
cooling water and partly (508,124 kJ/h) by refrigerant. Fig. 18
10 shows that the heat duty of reboilers in RD and DIST1 were com-
Optimized Case
9 pletely satisfied by heat integration with the background process
Base Case
streams (streams belonging to the process and not belonging to
8
condenser or reboiler of columns). Even the heat duty of reboiler
7 DIST2 was satisfied partially (746,076 kJ/h) by background
Stage

6 streams, however the remaining portion of the heat duty was sat-
isfied by using hot utility (high pressure steam). The condenser
5
duties of the columns (RD, DIST1 and DIST2) were mostly satisfied
4 by cold utility (cooling water). The data for the heat exchangers in
3 Fig. 18 is presented in Table A.5 (see Appendix).
For the heterogeneous catalyzed process (Fig. 19), the heating
2
and cooling requirements were 1,187,646 kJ/h (81.64 kW h/tonne
1 of biodiesel) and 1,747,410 kJ/h (120.12 kW h/tonne of biodiesel)
0 100000 200000 300000 400000 500000 600000 700000
respectively (3 heat exchangers, 2 utility coolers and 1 utility
)
Exergy Loss (kJ/hr
heater). The heating requirements were satisfied with high pres-
Fig. 15. Comparison of the stage–exergy loss profile of base case and optimized sure steam, while the cooling requirements were satisfied by
case design for RD2 column. cooling water (see Fig. 19). The data for the heat exchangers in

Fig. 16. Optimized process flowsheet for the homogeneous alkali-catalyzed process.
994 T. Poddar et al. / Applied Energy 185 (2017) 985–997

RD2 COOLER
DECANT
TRIOLEIN 2
OILFEED DEC-IN

HX1
PUMP2
FAME
MIXER VALVOUT
DISBT

MEOH
MEOHFEED
B1
3
MIXEDALC
1 VALVE GLY
PUMP1

HX2

PUMP3
DISTP

Fig. 17. Optimized process flowsheet for the heterogeneous catalyzed process.

23.3 0C 20 0C
C1

21.1 0C 20 0C
25 0C C2 20 0C
0
30.7 C COOLING WATER
C3

-24 0C C4 -25 0C
REFRIGERANT

0 157.3 0C 77.9 0C 72.3 0C

189.9 C E1 E3 E5 C3 25 0C
2 TO FINAL
112.2 0C 92.8 0C
112.3 0C E4 E6 E7 79.4 0C
FAME1 TO 1

101.9 0C C2 22.9 0C
CONDENSER DIST1
65.3 0C
85.9 0C E2 C1 64.5 0C
CONDENSER RD

5.7 0C C4 -19.8 0C
CONDENSER DIST2

147.4 0C 77.8 0C
189.7 0C H1 E3 E7 65.4 0C
REBOILER DIST2
105.6 0C
127.5 0C E1 E6 92.7 0C
REBOILER RD

102.5 0C E4 101.9 0C
REBOILER DIST1

60 0C E2 25 0C
OILFEED TO OIL2

60 0C E5 42.3 0C
MEOHFEED TO MEOH2

250 0C H1 249 0C
HP STEAM

Fig. 18. HEN diagram for the homogeneous alkali-catalyzed process.

Fig. 19 is presented in Table A.6 (see Appendix). It was interesting
to note that the utility requirements for the heterogeneous-
catalyzed process were larger than the homogeneous alkali cat-
alyzed process. A plausible reason for this could be the existence
of many downstream separation units in the alkali catalyzed pro-
cess, which results in more process streams available for heat
integration. The benefits of column optimization were also visible
in the usage of hot utility for both processes. The reboiler
temperature of DIST2 column (column with highest energy con-
sumption in alkali-catalyzed process) was reduced from
254.28 °C to 189.73 °C prompting the use of high pressure steam
as hot utility instead of a more expensive hot utility (such as hot
 T. Poddar et al. / Applied Energy 185 (2017) 985–997 995

22.7 0C 20 0C
C1

25 0C C2 20 0C
27.3 0C 20 0C
COOLING WATER

192.2 0C 156.1 0C 150.7 0C

227.8 0C E1 E2 E3 C2 25 0C
VALVOUT TO DEC-IN

118.9 0C C1 95.4 0C
CONDENSER RD2
213.3 0C
248.3 0C H1 E1 192.2 0C
REBOILER RD2

75 0C E2 25.3 0C
TRIOLEIN TO 2

75 0C E3 45.2 0C
MIXEDALC TO 3

250 0C H1 249 0C
HP STEAM

Fig. 19. HEN diagram for the heterogeneous-catalyzed process.

Table 4
Summary of main overall costs and economic indicators for both processes.

Item Alkali-catalyzed process Heterogeneous-catalyzed process

Total raw material cost ($/yr) 26,834,824 22,447,938
Total utility cost ($/yr) 62,312 65,858
A – plant capacity (tonnes/yr) 35444 35,394
B – Fixed Capital Investment (FCI) ($) 1,922,352 1,204,576
C – Total Capital Investment (TCI) ($) 2,210,704 1,385,263
D – annual production cost ($/yr) 30,044,533 25,215,927
E – annual product income ($/yr) 32,133,782 31,949,393
Return on investment (%) ((E D)/C) * 100 94.5 486
Payback period (yr) (B/(E D)) 0.9 0.2
Production cost per unit of biodiesel ($/kg) (D/A) 0.848 0.712

oil or fired heat). Similar trend was observed for RD2 column in
heterogeneous process, where the reboiler temperature was
reduced from 256.2 °C to 248.3 °C, enabling the use of high pres-
sure steam as hot utility.
7.4. Economic analysis
The costing used to calculate indirect costs as well as the break-
down of other important subsequent costs are shown in Table A.7
(see Appendix) and Table A.8 (see Appendix) for the homogeneous
alkali and heterogeneous process respectively. The annual produc-
tion cost is calculated by adding up the constituent costs as shown
in Tables A.9 (see Appendix) and A.10 (see Appendix) labeled (1) to
(11) [35]. After calculating the various economic indicators
described, the numbers obtained were compared between the
two production processes. The main costs and indicators are sum-
marized in Table 4. The TCI for the alkali-catalyzed process is 37.3%
larger than the heterogeneous-catalyzed process, which is at
$2,210,704 as compared to the $1,385,263. The annual production
cost for the alkali-catalyzed process ($30,044,533) was $5 million
higher than the heterogeneous-catalyzed process ($25,215,927).
For key economic indicators, the ROI for the heterogeneous process
was much higher than the alkali-catalyzed process with a 486%
return compared with 94.5%. The payback period for the heteroge-
neous process was also lower at 0.2 years for the heterogeneous
process as opposed to 0.9 years for the alkali-catalyzed process.
Table 4 shows the main associated costs and profitability indi-
cators of the alkali-catalyzed and heterogeneous-catalyzed reac-
tive distillation processes. From Table 4, it can be observed that
the FCI and TCI were higher for the alkali-catalyzed process than
the heterogeneous-catalyzed process. This is due to the use of
more downstream separation units for the alkali-catalyzed process
in comparison to the heterogeneous catalyzed process. Despite
both processes considering similar feedstock (soybean oil), and
the utility cost of the alkali-catalyzed process being lower than
that of the heterogeneous-catalyzed process; the annual produc-
tion cost for the heterogeneous catalyzed process is the lower of
the two. This is the result of additional raw materials, such as hex-
ane solvent and phosphoric acid, required in the alkali-catalyzed
process for downstream separation of biodiesel, glycerol and the
alkali catalyst. The aforementioned factors make the heteroge-
neous catalyzed process more cost-effective according to the prof-
itability indicators such as ROI, payback period, and unit
production costs (see Table 4).
From Tables A.9 (see Appendix) and A.10 (see Appendix), it can
be inferred that the cost of raw material has the most important
impact on the overall annual production cost. The cost of soybean
oil majorly dominates the raw material cost (see Table A.11). To
better understand the effect of the impact of the oil feedstock cost
996 T. Poddar et al. / Applied Energy 185 (2017) 985–997

on the profitability (i.e., ROI and payback period) of the biodiesel 0.9
production process, a sensitivity analysis was performed (see 0.8
Figs. 20 and 21). A similar analysis was performed using the biodie-
sel selling price of biodiesel. The results are shown in Figs. 22 and 0.7

23. The values for the ROI and the payback period were economi- 0.6

Payback Period (yr)

cally more attractive when the oil feedstock price was varied for
0.5
the heterogeneous process compared with the alkali-catalyzed
process. The breakeven feedstock price was found to be approxi- 0.4 Alkali-catalyzed
mately $0.66/kg and $0.78/kg for the alkali and heterogeneous cat- Heterogeneous
0.3
alyzed processes, respectively (see Fig. 20). On the other hand, as
0.2

0.1
2000
0
1500 0 0.5 1 1.5
Selling Price of Biodiesel ($/kg)
1000
Fig. 23. Plot depicting payback period vs. selling price of biodiesel.
500
ROI (%)

Alkali-catalyzed
0
shown in Fig. 22, the break-even price of biodiesel was found to
Heterogeneous
0 0.5 1 1.5 be approximately $0.73/kg and $0.60/kg for the alkali and hetero-
500 geneous catalyzed processes, respectively.

1000 8. Conclusion

1500 Cost of Oil Feedstock ($/kg) In this work, a techno-economic analysis of biodiesel produc-
tion by reactive distillation (alkali-catalyzed and heterogeneous-
Fig. 20. Plot depicting ROI vs. cost of oil feedstock.
catalyzed processes) was performed using a process simulation
approach. The simulations were followed by energy and economic
analyses in order to compare the both processes based on energy
1
and cost-effectiveness, while producing a fixed annual amount of
0.9 biodiesel. The energy analysis was performed by implementing a
0.8 column based optimization approach by studying the CGCC and
0.7
exergy loss profiles of the columns. This was followed by perform-
ing a heat integration analysis among the process streams and col-
Payback Period

0.6
umns. The economic analysis was undertaken by systematically
(yr)

0.5 calculating key capital and operating costs as well as profitability

Alkali-catalyzed
0.4 indicators; which were used for comparisons purposes. The annual
Heterogeneous
0.3 production cost and capital expenditure were found to be lower for
the heterogeneous catalyzed process. From the profitability analy-
0.2
sis, it can be concluded that the heterogeneous-catalyzed process
0.1 of biodiesel production is more economically advantageous than
0 the alkali-catalyzed process. This is due to lower capital invest-
0 0.2 0.4 0.6 0.8 ment, lower annual production cost, higher return on investment
Cost of Oil Feedstock ($/kg) and shorter payback period. Furthermore, it was found that for a
processing plant with a capacity of 35.4 kilotonnes/year, and a
Fig. 21. Plot depicting payback period vs. cost of oil feedstock.
price of $0.60/kg for the oil feedstock (keeping the remaining
parameters at their default values), the biodiesel unit production
cost were $0.716/kg and $0.858/kg for heterogeneous and alkali-
catalyzed processes respectively. These prices were close to the
2000
current selling price of biodiesel; which ranges between $0.79/kg
and $1.00/kg; thereby validating the correctness of the present
1500
analysis.
Alkali-catalyzed
1000
Heterogeneous
Acknowledgement
500
ROI (%)

The authors would like to thank The Petroleum Institute, Abu

0 Dhabi, for the financial support received during the course of this
0 0.5 1 1.5
work.
500

1000 Appendix A. Supplementary material

1500 Supplementary data associated with this article can be found, in

Selling Price of Biodiesel ($/kg)
the online version, at http://dx.doi.org/10.1016/j.apenergy.2015.
Fig. 22. Plot depicting ROI vs. selling price of biodiesel. 12.054.
 T. Poddar et al. / Applied Energy 185 (2017) 985–997 997

References [22] Gaurav A, Leite ML, Ng FTT, Rempel GL. Transesterification of triglyceride to
fatty acid alkyl esters (biodiesel): comparison of utility requirements and
capital costs between reaction separation and catalytic distillation
[1] Zhang Y, Dube MA, McLean DD, Kates M. Biodiesel production from waste
configurations. Energy Fuel 2013;27:6847–57.
cooking oil: 1. Process design and technological assessment. Bioresour Technol
[23] Boon-anuwat N-n, Kiatkittipong W, Aiouache F, Assabumrungrat S. Process
2003;89:1–16.
design of continuous biodiesel production by reactive distillation: comparison
[2] Banchero M, Kusumaningtyas RD, Gozzelino G. Reactive distillation in the
between homogeneous and heterogeneous catalysts. Chem Eng Process
intensification of oleic acid esterification with methanol – A simulation case-
Process Intensif 2015;92:33–44.
study. J Ind Eng Chem 2014;20:4242–9.
[24] Campanelli P, Banchero M, Manna L. Synthesis of biodiesel from edible, non-
[3] Banchero M, Gozzelino G. Nb2O5-catalyzed kinetics of fatty acids esterification
edible and waste cooking oils via supercritical methyl acetate
for reactive distillation process simulation. Chem Eng Res Des
transesterification. Fuel 2010;89:3675–82.
2015;100:292–301.
[25] Lavric V. Process design, integration and optimisation: advantages, challenges
[4] West AH, Posarac D, Ellis N. Assessment of four biodiesel production processes
and drivers. In: Klemeš JJ, editor. Handbook of process integration
using HYSYS.Plant. Bioresour Technol 2008;99:6587–601.
(PI). Woodhead Publishing; 2013. p. 79–125.
[5] Zhang Y, Dube MA, McLean DD, Kates M. Biodiesel production from waste
[26] Shahandeh H, Ivakpour J, Kasiri N. Internal and external HIDiCs (heat-
cooking oil: 2. Economic assessment and sensitivity analysis. Bioresour
integrated distillation columns) optimization by genetic algorithm. Energy
Technol 2003;90:229–40.
2014;64:875–86.
[6] Haas MJ, McAloon AJ, Yee WC, Foglia TA. A process model to estimate biodiesel
[27] Chen T, Zhang B, Chen Q. Heat integration of fractionating systems in para-
production costs. Bioresour Technol 2006;97:671–8.
xylene plants based on column optimization. Energy 2014;72:311–21.
[7] Harding KG, Dennis JS, von Blottnitz H, Harrison STL. A life-cycle comparison
[28] Demirel Y. Sustainable operations for distillation columns. Chem Eng Process
between inorganic and biological catalysis for the production of biodiesel. J
Tech 2013;1:1005.
Clean Prod 2008;16:1368–78.
[29] Nguyen N, Demirel Y. Retrofit of distillation columns in biodiesel production
[8] Marchetti JM, Miguel VU, Errazu AF. Techno-economic study of different
plants. Energy 2010;35:1625–32.
alternatives for biodiesel production. Fuel Process Technol 2008;89:740–8.
[30] Demirel Y. Retrofit of distillation columns using thermodynamic analysis. Sep
[9] Marchetti JM, Errazu AF. Technoeconomic study of supercritical biodiesel
Sci Technol 2006;41:791–817.
production plant. Energy Convers Manage 2008;49:2160–4.
[31] Demirel Y. Using the second law: thermodynamic analysis. In: Demirel Y,
[10] Apostolakou AA, Kookos IK, Marazioti C, Angelopoulos KC. Techno-economic
editor. Nonequilibrium thermodynamics. Amsterdam: Elsevier Science B.V.;
analysis of a biodiesel production process from vegetable oils. Fuel Process
2007. p. 155–274.
Technol 2009;90:1023–31.
[32] Dhole VR, Linnhoff B. Distillation column targets. Comput Chem Eng
[11] Sharma S, Rangaiah GP. Multi-objective optimization of a bio-diesel
1993;17:549–60.
production process. Fuel 2013;103:269–77.
[33] Turton R, Bailie RC, Whiting WB, Shaeiwitz JA, Bhattacharyya D. Analysis,
[12] Patle DS, Sharma S, Ahmad Z, Rangaiah GP. Multi-objective optimization of
synthesis and design of chemical processes. 4th ed. Pearson Educational
two alkali catalyzed processes for biodiesel from waste cooking oil. Energy
International; 2013.
Convers Manage 2014;85:361–72.
[34] Seider WD, Seader JD, Lewin DR. Product and process design principles:
[13] Patle DS, Ahmad Z, Rangaiah GP. Plantwide control of biodiesel production
synthesis, analysis and evaluation, (with CD). John Wiley & Sons; 2009.
from waste cooking oil using integrated framework of simulation and
[35] Hunpinyo P, Narataruksa P, Tungkamani S, Pana-Suppamassadu K, Chollacoop
heuristics. Ind Eng Chem Res 2014;53:14408–18.
N. Evaluation of techno-economic feasibility biomass-to-energy by using
[14] Cossio-Vargas E, Hernandez S, Segovia-Hernandez JG, Cano-Rodriguez MI.
ASPEN PlusÒ: a case study of Thailand. Energ Procedia 2013;42:640–9.
Simulation study of the production of biodiesel using feedstock mixtures of
[36] Soybean Oil Monthly Price; 2015. <http://www.
fatty acids in complex reactive distillation columns. Energy 2011;36:6289–97.
indexmundi.com/commodities/?commodity=soybean-oil>.
[15] Kiss AA. Heat-integrated reactive distillation process for synthesis of fatty
[37] Regional Contract Methanol Prices. Methanex Corporation; 2015. <https://
esters. Fuel Process Technol 2011;92:1288–96.
www.methanex.com/our-business/pricing>.
[16] Kiss AA, Bildea CS. A review of biodiesel production by integrated reactive
[38] Alternative Fuel Price Report; 2015. <http://www.afdc.energy.gov/fuels/prices.
separation technologies. J Chem Technol Biotechnol 2012;87:861–79.
html>.
[17] Machado GD, Aranda DAG, Castier M, Cabral VF, Cardozo-Filho L. Computer
simulation of fatty acid esterification in reactive distillation columns. Ind Eng
Chem Res 2011;50:10176–84. Glossary
[18] Machado GD, Pessoa FLP, Castier M, Aranda DAG, Cabral VF, Cardozo-Filho L.
Biodiesel production by esterification of hydrolyzed soybean oil with ethanol CGCC: Column Grand Composite Curve
in reactive distillation columns: simulation studies. Ind Eng Chem Res
PNMTC: Practical Near Minimum Thermodynamic Condition
2013;52:9461–9.
ROI: Return on Investment
[19] Mueanmas C, Prasertsit K, Tongurai C. Transesterification of triolein with
FAME: Fatty Acid Methyl Ester
methanol in reactive distillation column: simulation studies. Int J Chem React
Eng 2010;8:1542–6580. HEN: Heat Exchanger Network
[20] Prasertsit K, Mueanmas C, Tongurai C. Transesterification of palm oil with TIIC: Total Installation Cost
methanol in a reactive distillation column. Chem Eng Process Process Intensif TIC: Indirect Capital Cost
2013;70:21–6. DPC: Direct Permanent Cost
[21] Simasatitkul L, Siricharnsakunchai P, Patcharavorachot Y, Assabumrungrat S, FCI: Fixed Capital Investment
Arpornwichanop A. Reactive distillation for biodiesel production from soybean TCI: Total Capital Investment
oil. Korean J Chem Eng 2011;28:649–55.