chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

journal homepage: www.elsevier.com/locate/cherd

Process simulation and economic analysis of biodiesel

production processes using fresh and waste vegetable oil
and supercritical methanol

Soojin Lee, Dusko Posarac, Naoko Ellis ∗

Department of Chemical and Biological Engineering, University of British Columbia, Vancouver, BC, Canada V6T 1Z3

a b s t r a c t

Three continuous biodiesel processes with production capacity of 40,000 tonne/yr, including a conventional alkali-
catalyzed process using both fresh and waste vegetable oil and a supercritical methanol process using waste vegetable
oil as the raw material, were simulated in HYSYS. In order to improve the simulation accuracy, the properties of tri-
olein, a model compound of vegetable oil, were re-evaluated. The normal boiling point of triolein was experimentally
determined by thermogravimetric analysis and further incorporated in HYSYS simulation, which resulted in improve-
ments in the values of specific heat capacity, mass density, and viscosity. Process economics were analyzed using
Aspen In-Plant Cost Estimator. The alkali-catalyzed process using fresh vegetable oil had the lowest total capital
investment, but the supercritical process was the most economically feasible overall, providing a lower manufactur-
ing cost and higher net present value and a discounted cash flow rate of return. Sensitivity analyses of net present
value were conducted using four parameters including oil feedstock costs, glycerol credit, biodiesel selling prices,
and interest rates. Based on the analyses, prediction equations of net present value were developed.
© 2011 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Keywords: Supercritical biodiesel production; Process simulation; Triolein; Economic assessment; Waste vegetable
oil; Energy consumption

1. Introduction alkali-catalysts. Feedstocks for biodiesel are mainly food-

grade vegetable oils. However, recent research has moved
Steady increase in prices of petroleum-based fuels and grow- toward alternative feedstocks such as waste vegetable oils
ing environmental concerns are boosting public attention to (WVOs) and non-edible vegetable oils because the use of food-
alternative fuels. Moreover, worldwide energy demand has grade oils is becoming less economically viable as a result of
steadily increased, widening the gap between the supply and the increasing prices of crops as shown in Fig. 1.
demand of petroleum oil, and raising concerns about the However, it is well known that the conventional base cat-
depletion of fossil fuels. Biodiesel has drawn attention as alytic processes are inefficient when using low-cost feedstocks
an alternative fuel, especially as a substitute to petroleum- such as waste vegetable oils due to the presence of large
derived diesel. amounts of free fatty acids (FFAs) and water. These impurities
Biodiesel is defined as a mixture of fatty acid alkyl esters are known to consume the alkali catalysts via saponification
which are commonly produced from triglycerides and alco- reaction, thereby decreasing biodiesel yields. Moreover, a mix-
hol through transesterification reaction in the presence of ture of soaps, biodiesel and un-reacted compounds causing

Abbreviations: ANOVA, analysis of variance; ASME, American society of mechanical engineering; B100, 100% pure biodiesel fuel; CCD,
central composite design; CSTR, continuous stirred tank reactor; DCFROR, discounted cash flow rate of return; DPP, discounted payback
period; EOS, equation of state; FFA, free fatty acid; FVO, fresh vegetable oil; G&A, general & administrative; MARR, minimum acceptable
rate of return; MOC, material of construction; NPV, net present value; NRTL, non-random two liquid model; PFR, plug flow reactor; Prob,
probability; RSM, response surface model; TGA, thermogravimetric analysis; UNIFAC, universal functional activity coefficient; WVO, waste
vegetable oil.
∗
Corresponding author. Tel.: +1 604 822 1243; fax: +1 604 822 6003.
E-mail addresses: [email protected] (S. Lee), [email protected] (D. Posarac), [email protected] (N. Ellis).
Received 26 September 2010; Received in revised form 28 April 2011; Accepted 22 May 2011
0263-8762/$ – see front matter © 2011 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.cherd.2011.05.011
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642 2627

tage of a supercritical process. Another investigation on the

Nomenclature supercritical biodiesel production from WVO was reported by
Kasteren and Nisworo (2007). They used propane as a co-
Pc critical pressure (kPa) solvent to alleviate harsh supercritical reaction conditions.
Tb boiling point (◦ C) The required selling prices for biodiesel were predicted to be
Tc critical temperature (◦ C) US$ 0.17/l, US$ 0.24/l and US$ 0.52/l at the plant capacities of
Vc critical volume (m3 /kmol) 125,000, 80,000 and 8000 tonne biodiesel/yr, respectively.
ω acentricity Most biodiesel economic studies are based on process sim-
ulation, which is regarded as one of the best ways to conduct
the economic assessment. Triolein is routinely used as a
emulsification during a washing process makes it difficult to model compound of vegetable oils (Zhang et al., 2003; Kasteren
obtain biodiesel of high quality (Marulanda et al., 2010). and Nisworo, 2007; West et al., 2008; Glisic et al., 2009; Lim
Supercritical methanol transesterification is one of the et al., 2009) because it composes around 40–80% of fatty acids
ways to utilize those feedstocks. Kusdiana and Saka (2004) in a variety of vegetable oils such as rapeseed, canola, olive,
reported that almost complete conversion was achieved using palm and peanut oils (Pinnarat and Savage, 2008). Despite the
supercritical methanol and vegetable oil containing FFAs up significance, the chemical properties of triolein have not been
to 30 wt%. In addition, the yield of methyl esters from the thoroughly understood. The reported normal boiling point
supercritical reaction was not affected by up to 30 vol% water and critical properties of triolein vary significantly by differ-
content in the feedstock. In comparison with the supercritical ent estimation methods and process modeling programs, as
method, vegetable oil with 20 wt% FFAs resulted in only 35% shown in Table 1. The use of comprehensive properties of the
conversion in the alkali-catalyzed method. major components results in accurate and reliable simulation
West et al. (2008) evaluated the costs involved in produc- results.
ing 8000 tonne/yr of biodiesel from waste cooking oil for Among the properties of triolein, the normal boiling point is
four processes: a homogeneous alkali-catalyzed process; a of major interest since it is used in predicting other properties
homogeneous acid catalyzed process; a heterogeneous acid such as critical temperature and pressure. Goodrum and Geller
catalyzed process; and a supercritical process. The results (2002) applied thermogravimetric analysis (TGA) method to
showed that the supercritical process was the second most measure the normal boiling point of medium- and long-chain
profitable process next to the heterogeneous catalyzed pro- triglycerides, namely trilaurin (C12:0), trimyristin (C14:0), tri-
cess. Lim et al. (2009) conducted economic analysis of the palmitin (C16:0) and tristearin (C18:0). The results were in good
processes for supercritical biodiesel production from rape- accordance with the Clausius-Clapeyron model and with pre-
seed oil and compared its profitability to an alkali-catalyzed viously published data at low pressures (Perry et al., 1949).
process. They found that the total manufacturing cost of the Ceriani and Meirelles (2004) proposed a group contribution
supercritical process was lower than the alkali-catalyzed pro- method to estimate vapour pressures of fatty compounds. The
cess due to the absence of catalyst and higher by-product estimation model includes temperature and a few parameters
(glycerol) credit. Despite the higher total capital cost of the to take into account various molecular groups. The parameters
supercritical process, its lower manufacturing cost resulted were obtained by regressions of more than 1300 experimental
in shorter payout time than the alkali-catalyzed process. The values of fatty compounds (saturated and unsaturated fatty
faster payout time of the supercritical processes, even without acids, fatty esters, fatty alcohols, monoglycerides and triglyc-
using waste vegetable oil feedstocks, indicates the advan- erides) gathered from open literature.

Fig. 1 – The prices of vegetable oils from Oct 2000 to Jan 2010. * Average from October to the following September.
Source: USDA, 2010a.
2628 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642

Table 1 – Comparison of available properties of triolein.

HYSYS Aspen VMGsim Tang et al. (2006) Glisic et al. (2007) Weber et al. (1999)
a
◦
Tb ( C) 606.8 846.85 541.7 879.9 – –
Tc (◦ C) 680.9 1367 681 954.1a 704.73b 673.92c
Pc (kPa) 360.2 470 360.2 360.2a 334b 468.20c
Vc (m3 /kmol) 3.09 3.09 3.09 – – –
ω 1.686 – 1.69 1.6862d 1.978b 1.686

a
Estimated by the method of Dohrn and Brunner (1991, 1994).
b
Estimated by the method of Constantinou and Gani (1994), Constantinou et al. (1995).
c
Estimated by the method of Ambrose (1978) and Reid et al. (1987).
d
Estimated according to Han and Peng (1993).

In this study, the normal boiling point of triolein was spheric pressure. However, compared to the values listed in
estimated using both TGA and the group contribution meth- Table 1, the group contribution method estimated the boiling
ods, and the results were compared with those from the point much closer to the experimental results in this study.
literature and process simulators. The normal boiling point
obtained from the TGA was subsequently incorporated in
the process simulator, HYSYS, during the development of a 2.2. Properties of triolein
simulation model of a supercritical biodiesel production pro-
cess. The model was then economically assessed to estimate The experimental TGA boiling point was then used in HYSYS
total capital investment and manufacturing costs of biodiesel along with the molar mass (885.4 g/mol), ideal liquid density
and evaluate the profitability. To assist in determining the (916 kg/m3 ) and UNIFAC structure to estimate critical and ther-
desirability of the supercritical process, the results were also modynamic properties of triolein. Temperature dependent
compared with those of homogeneous alkali-catalyzed pro- properties such as vapour enthalpy, vapour pressure and Gibbs
cesses using both fresh and waste vegetable oils. All economic free energy were also estimated. This triolein was named
studies conducted in here were first done using Aspen In-Plant TGA Triolein to make distinction from the triolein built in
Cost Estimator. HYSYS component library, named HYSYS Triolein. There were
discrepancies observed between TGA and HYSYS Triolein, as
2. Triolein a model compound of vegetable shown in Table 3, in specific heat capacity, mass density and
oils viscosity.
As shown in Table 3, the specific heat capacity and viscos-
2.1. Thermogravimetric analysis ity of TGA Triolein were 12 and 19 times higher than HYSYS
Triolein, respectively. The mass density of TGA Triolein was
The details of the methods of thermogravimetric analysis of 58% of that of HYSYS Triolein. Overall, the properties of TGA
triglycerides are described in Goodrum and Geller (2002). Triolein were found to be much closer to the properties of
Fig. 2 shows a weight loss curve of a sample consisted of real canola oil. Especially, the mass density of triolein was
4.91 mg of triolein and 1.08 mg of alumina powder. The sam- much improved as considering the common knowledge of oil
ple weight remained constant until the temperature reached having lower density than water. These indicate that improve-
370 ◦ C, and then rapidly decreased after 400 ◦ C. After the evap- ments can be achieved in a process simulation by replacing
oration of triolein, a second constant line appeared at the HYSYS Triolein with TGA Triolein. As an example, the dif-
20% of the initial sample weight, indicating that only the alu- ference in specific heat capacity of triolein would affect the
mina powder added remained in the pan. All five replicate energy consumptions in a supercritical process where the raw
runs showed the same single rapid and definite weight-loss materials need to be heated up to high temperatures (>280 ◦ C).
curve without any secondary transition. This fact indicates In addition, due to the existence of unreacted triolein, the
that there was minimal thermal decomposition during heat- performance of the biodiesel purification steps may also be
ing since the weight curve would have had multiple transition altered.
points if the analyzed material were decomposed into small
molecules. The average normal boiling point obtained from
the replicate runs was 412.8 ◦ C with a standard deviation of 3. Process simulation
1.08 ◦ C. Table 2 summarizes the normal boiling points of tri-
olein and tripalmitin estimated experimentally and by the 3.1. Chemical components, thermodynamic model and
group contribution method. plant capacity
The boiling points of triolein and tripalmitin estimated by
the group contribution method were approximately 30–40 ◦ C HYSYS Version 2006.5 was used to conduct the simulation.
higher than the experimental results. The main reason for the TGA Triolein (C57 H98 O6 ) developed in Section 2 was used as a
differences appears to be generated from the characteristic of model compound to represent canola oil. Values for density,
the group contribution method itself. Since the group contri- boiling point, critical temperature, pressure, and volume were
bution method was developed to be used for a broad range of 916.0 kg/m3 , 412.8 ◦ C, 588.8 ◦ C, 1203 kPa, and 1.346 m3 /kmol,
fatty compounds, obtaining the boiling point of a particular respectively. Fresh canola oil was modeled as 100% TGA Tri-
compound may slightly deviate from the true value. Further- olein, and waste canola oil was modeled as a mixture of 94 wt%
more, the parameters in the method were obtained from the TGA Triolein and 6 wt% oleic acid (C18 H34 O2 ). Consequently,
regression of the experimental data in literature, which are methyl oleate (C19 H36 O2 ) was considered as the product of
very rare in case of the boiling point of triglycerides at atmo- esterification and transesterification reactions.
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642 2629

Fig. 2 – Weight loss curve of triolein (C18:1) measured in TGA.

Table 2 – Comparison of boiling point of tripalmitin and triolein obtained from TGA and group contribution method.
Compound TGA (◦ C) (this TGA (◦ C) (Goodrum and Group contribution method (◦ C)
study) Geller, 2002) (Ceriani and Meirelles, 2004)

Tripalmitin (C16:0) 409.3 399.08 449.6

Triolein (C18:1) 412.8 N/A 432.2

Non-random two liquid (NRTL) model was selected as a process, named SC-WVO, was a supercritical process using
property package due to the presence of polar compounds. waste canola oil. The process flowsheet and operating con-
Some interaction parameter coefficients between components ditions for Alkali-FVO and Alkali-WVO processes referred to
were not available in HYSYS library. These parameters were those reported by Zhang et al. (2003) and West et al. (2008). The
estimated using UNIFAC LLE model. Since an activity coef- flowsheets and details of major material streams of the pro-
ficient based model such as NRTL is not recommended to cesses are presented in Figs. 3 and 4. The reaction conditions
be used at pressures greater than 1000 kPa, Peng-Robinson are summarized in Table 4.
equation of state (EOS) was used in process streams and unit
operators wherever having pressures >1000 kPa. 3.2.1. SC-WVO process
The plant annual capacity was specified at 40,000 As stated previously, a supercritical process is not sensitive
tonne/year of biodiesel production based on 8000 operating to the presence of FFAs and water in feedstocks. Therefore,
hours per year. waste vegetable oil was chosen as the raw material for SC-
WVO process. SC-WVO process basically consisted of 4 steps:
3.2. Process design heating and pressurizing of raw materials, transesterification
and esterification reactions, methanol recovery and biodiesel
Three biodiesel production processes were simulated. The first purification. The flowsheets are presented in Fig. 5.
process, named Alkali-FVO, was an alkali-catalyzed process
using fresh canola oil as the feedstock. The second process, 3.2.2. Heating and pressurizing of raw materials
named Alkali-WVO, was an alkali-catalyzed process with an Methanol (stream 101) and WVO (stream 103) were pressur-
acid-catalyzed pre-treatment step of waste canola oil. The last ized to the reaction pressure (19 MPa) by P-101 and P-102,

Table 3 – Comparison of the properties of Hysys triolein, TGA triolein and canola oil.
Specific heat capacity (kJ/kg ◦ C) Mass density (kg/m3 ) Viscosity (cP)
a
HYSYS Triolein 0.131 1571 1.765
TGA Trioleina 1.58 909.8 33.19
Canola oilb 1.910–1.916 914–917 71.5–71.7

a
Properties at 25 ◦ C and 1 atm.
b
Przybylski (2008).
2630 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642

Fig. 3 – Flow diagram and stream table of Alkali-FVO process.

respectively. Then each pressurized stream flowed through 3.2.4. Methanol recovery
heat exchangers (E-102 and E-101) where the heat of reac- A significant portion of methanol in the reactor effluent
tion product stream heated the waste oil to 293.9 ◦ C and then stream can be separated out as a vapour in a flash evaporator
the methanol to 211.0 ◦ C. The raw material streams were then (V-100) by adjusting the pressure or temperature. VLV-100 was
mixed together in a mixer (MIX-100) and brought to the desired used to depressurize stream 106B from 19 to 0.2 MPa so that
temperature (350 ◦ C) by E-103. most of methanol evaporates whereas the other components
remain mostly in the liquid phase. Using this procedure, 89.6%
of methanol in the feed stream was separated into stream 109.
3.2.3. Biodiesel production in a plug flow reactor (PFR)
Then, a distillation column with 4 theoretical stages was used
For continuous biodiesel production, a PFR was selected to
to further separate 99.93% of methanol in stream 108. The
carry out transesterification and esterification reactions. In
recovered methanol was then recycled and mixed with the
this study, the reaction conditions and biodiesel yield were
fresh methanol feed (stream 101).
obtained from Bunyakiat et al. (2006) where the reaction was
carried out in a continuous method using a 5.5 m in length
with 3/8-in. (9.525 mm) o.d. and 0.035-in. (0.889 mm) thickness
316SS tubing. By matching the experiment results shown in 3.2.5. Glycerol separation
Table 5, 96% conversion of triolein was simulated in the PFR at After methanol recovery in the flash drum and the distillation
350 ◦ C and 19 MPa with 800 s of reaction time. The molar ratio column, the bottom stream was cooled down to 25 ◦ C and fed
of methanol to oil was 24 and the reactor was considered to to a decanter. Two liquid phases were formed: a glycerol-rich
operate isothermally. phase and a methyl oleate-rich phase. The separation result

Table 4 – Summary of reactions involved in simulation models.

Alkali-FVO Alkali-WVO SC-WVO

Reactions Transesterification Esterification/transesterification Simultaneous transesterification and

esterification
Catalyst NaOH H2 SO4 /NaOH –
Temperature (◦ C) 60 70/60 350
Pressure (kPa) 400 400/400 19000
Reaction time (min) 60 60/60 13.3
Methanol:oil (molar) 6:1 6:1/6:1 24:1
Conversion (%) 95 100/95 96a
Reactor type CSTR CSTR/CSTR PFR

a
For triolein. The conversion of oleic acid is 100%.
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642 2631

Fig. 4 – (a) Flow diagram and stream table of pre-treatment in Alkali-WVO process. (b) Flow diagram and stream table of
Alkali-WVO process after pre-treatment.

showed that 99.7% of glycerol in stream 109A was drained out ified in EN 14214, further purification of stream 204 in a
with the denser phase in this unit. distillation column was necessary. Since the normal boiling
temperature of methyl oleate is as high as 359 ◦ C, the pressure
3.2.6. Biodiesel purification inside T-102 was set at as low as 3.0–5.0 kPa in order to avoid
Stream 204 consisted of 96.2 wt% methyl oleate and the bal- thermal decompositions of methyl esters. The number of the-
ance was mostly triolein. To satisfy the requirement of the oretical stages used was 7. The distillate (stream 206) had a
maximum triglyceride content in biodiesel (≤0.20 wt%) spec- purity of 99.8 wt% methyl oleate.
2632 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642

Fig. 5 – Flow diagram and stream table of SC-WVO process.

Table 5 – Raw materials, reaction conditions, yields and reactor types for supercritical biodiesel production.
Oil T (◦ C) P (MPa) Methanol/oila  (min) B/Cb Yield (%) Ref.

Sunflower 400 20 40 40 B 96 Madras et al. (2004)

Rapeseed 350 43 42 4 B 95 Kusdiana and Saka (2001)
Jatropha 320 8.4 43 4 B 100 Hawash et al. (2009)
Soybeen 310 32 40 25 C 96 He et al. (2007)
Castor/linseed 350 20 40 40 B 98 Varma and Madras (2007)
Palm 350 19 42 7 C 96 Bunyakiat et al. (2006)

a
Molar ratio.
b
B-batch or C-continuous.

3.3. Energy consumption than Alkali-FVO due to the high energy consumption of
the methanol distillation columns. For instance, the first
Energy consumptions of the three process simulations were methanol distillation column (T-100) in the pre-treatment
compared in Table 6. step in Fig. 4a consumes 2279 kW, nearly half of the total
Alkali-FVO had the lowest energy consumption among energy consumption. In comparison with the alkali-catalyzed
the three processes. Alkali-WVO consumed more energy processes, the supercritical processes had a large energy
requirement for pumping and heating raw material to achieve
the harsh reaction conditions. Despite this fact, the energy
Table 6 – Energy consumption (kW) of the three process
simulation models. consumption in SC-WVO process was less than that of the
Alkali-WVO process. Since a large portion of methanol was
Alkali-FVO Alkali-WVO SC-WVO
separated in the flash evaporator and the lower methanol
Pumps 1.043 1.688 87.73 to oil molar ratio was used in this process, the methanol
Heaters 83.45 34.05 1429 flow rate to the distillation column is lower, resulting in
Distillation columns (Reboiler) the total energy consumption comparable to the Alkali-FVO
Methanol 664 2279 (T − 100) 792.7 process.
recovery
873.8 (T − 101)

Biodiesel 2069 2033 1618 4. Economic assessment

purification
Glycerol 31.92 36.78 – Previous studies on economic estimations were based on the
purification overall factor method of Lang or bare-module concept intro-
Total 2349 5258 3927
duced by Guthrie (Zhang et al., 2003; Kasteren and Nisworo,
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642 2633

Table 7 – The general specifications for capital costs of the three biodiesel production processes.
Alkali-FVO Alkali-WVO SC-WVO

Process Description Proven process Proven process New process

Process Complexity Typical Typical Highly complex
Process Control Digital Digital Digital
Project Location North America North America North America
Currency U.S. dollar U.S. dollar U.S. dollar
Project Type Grass roots Grass roots Grass roots
Contingency (%) 18.0 18.0 32.5
Estimated Start date of Basic Engineering 2009/01/01 2009/01/01 2009/01/01
Soil condition around site Soft clay Soft clay Soft clay
Pressure vessel design code ASMEa ASMEa ASMEa

a
American Society of Mechanical Engineers, Section VIII, Division 1.

2007; You et al., 2008; West et al., 2008; Lim et al., 2009; Santana The specific types and the material of construction (MOC)
et al., 2010). These methods provide an approximate cost data of process equipments were defined and sized according to
and have relatively poor accuracies of estimation (Turton, principles outlined in the literatures (Luyben, 2002; Seider
2003). We were not able to find economic analysis of a biodiesel et al., 2004; Chopey, 2004). The details are shown in Tables 8–10.
plant using a software package process evaluation in the open
literature. As a result, we have decided to use Aspen In-Plant 4.1. Investment parameters
Cost Estimator, since it has been used for over 30 years in com-
mercial plants and engineering designs, and provides more The specified investment parameters are summarized in
accurate estimation (Seider et al., 2004). Aspen In-Plant Cost Table 11.
Estimator provides specifications for detailed design, estima- Table 12 gives the prices of raw materials and chemicals
tion and economic data, allowing quick modifications of the used in this study. The list of the utilities is shown in Table 13.
process equipment and sensitivity analysis. The general spec- Since the temperature of high-pressure saturated steam is not
ifications for capital costs used in this study are summarized sufficient to heat up the reactant streams to reaction temper-
in Table 7. ature (350 ◦ C) in SC-WVO process, hot temperature heating oil

Table 8 – Equipment summary for Alkali-FVO process.

Pumps P-101 P-102 P-103 P-202

Type Centrifuge Centrifuge Centrifuge Centrifuge

Efficiency 75% 75% 75% 75%
MOC 316SS CS CS CS
Flow (kg/h) 617.8 5250 574.4 5868
Press in (kPa) 101.3 101.3 20.0 30.0
Press out (kPa) 400.0 400.0 400.0 120.0
Driver power (kW) 0.0858 0.6384 0.1033 0.2154

Heat exchangers E-101 E-102

Type Double pipe Fixed head

Area (m2 ) 1.0 5.97
MOC CS CS
Duty (kW) 83.45 307.8
Shell in (◦ C/kPa) 25.3/400 142.8/30
Tube in (◦ C/kPa) 164/690 23.8/345

Vessels/reactors CRV-100 CRV-101 X-100

Type Agitated reactor Agitated reactor L-L separator

Orientation Vertical Vertical Horizontal
MOC 316SS 316SS 316SS
T (◦ C) 60.0 161.2 161.2
P (kPa) 400.0 101.3 101.3
Residence time (min) 60 40 60
Diameter (m) 1.847 0.720 0.802
Height/Length (m) 5.541 2.160 2.406

Tower T-101 T-102 T-103 T-104

Application Distillation Extraction Distillation Distillation

# of actual stages 9 6 9 8
Type of trays/Packing materials Valve Sieve Valve 1 in. Pall rings
MOC CS CS CS CS
Diameter (m) 0.90 0.602 2.440 0.3048
Height (m) 6.858 9.144 6.858 4.572
2634 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642

Table 9 – Equipment summary for Alkali-WVO process.

Pumps P-101 P-102 P-103 P-104 P-105 P-106 P-107

Type Centrifuge Centrifuge Centrifuge Centrifuge Centrifuge Centrifuge Centrifuge

Efficiency 75% 75% 75% 75% 75% 75% 75%
MOC 316SS CS CS CS SS316 CS CS
Flow (kg/h) 173.1 1167 5275 5256 560.6 542.9 5817
Press in (kPa) 101.3 28.0 101.3 200.0 101.3 20.0 30.0
Press out (kPa) 400.0 400.0 400.0 400.0 400.0 400.0 120.0
Driver power (kW) 0.0199 0.2073 0.6405 0.4242 0.0778 0.1046 0.2141

Heat exchangers E-101 E-102 E-103

Type Float head Double pipe Fixed head

Area (m2 ) 35.18 1.0 6.47
MOC 316SS CS CS
Duty (kW) 849.58 34.05 358.11
Shell in (◦ C/kPa) 25.3/400 46.4/400 156/30
Tube in (◦ C/kPa) 70/400 164/690 23.8/345

Vessels/Reactors CRV-100 CRV-101 CRV-102 X-102

Type Agitated reactor Agitated reactor Agitated reactor L-L separator

Orientation Vertical Vertical Vertical Horizontal
MOC 316SS 316SS 316SS 316SS
T (◦ C) 70.0 60.0 157.1 157.1
P (kPa) 400.0 400.0 101.3 101.3
Residence time (min) 60 60 40 60
Diameter (m) 1.863 1.830 0.712 0.904
Height/length (m) 5.589 5.490 2.136 2.712

Tower T-100 T-101 T-102 T-103 T-104 X-100

Application Distillation Distillation Extraction Distillation Distillation Extraction

# of actual stages 8 9 6 9 8 6
Type of trays/packing materials Valve Valve Sieve Valve 1 in. Pall ring Sieve
MOC 316SS CS CS CS CS CS
Diameter (m) 1.524 0.9144 0.613 2.286 0.3048 0.646
Height (m) 6.096 6.858 9.144 6.858 4.572 9.144

was used to provide heat in the process wherever the operating and 75.4% of the total manufacturing costs of Alkali-FVO,
temperatures were greater than 254 ◦ C. Alkali-WVO, and SC-WVO processes, respectively. Due to the
usage of fresh vegetable oil feedstock, it was especially high
4.2. Project evaluation in Alkali-FVO process. The other two processes using waste
vegetable oil had approximately 42% lower oil feed costs.
4.2.1. Total capital investment The total manufacturing cost of Alkali-WVO process was
Table 14 shows the total capital costs for the three production $35.0 million, which was the second highest among the pro-
processes. Overall, the purchased equipment cost of SC-WVO cesses. Compared to Alkali-FVO, the cost for raw materials
was the highest. The supercritical process had the highest except for the oil feed was higher because of the usage of larger
costs of the reactor, pumps and heat exchangers due to the amount of methanol, sulphuric acid catalyst and glycerol for
high operating pressure and temperature. washing solution in the liquid–liquid extraction column in the
The total direct costs of Alkali-FVO, Alkali-WVO, and SC- pre-treatment process. The utility, maintenance and indirect
WVO processes were estimated as $3.594, $5.336, and $5.425 manufacturing costs of Alkali-WVO were also slightly higher
million, respectively. The largest number of pieces of equip- than Alkali-FVO due to the complexity of the process. How-
ment in Alkali-WVO process generated the highest costs for ever, using waste vegetable oil feedstock saved $10.8 million
equipment setting, civil, steel, instrumentation, electrical, per year in the total manufacturing cost of Alkali-WVO. SC-
insulation and paint. The total capital investment for Alkali- WVO process showed lower total manufacturing costs than
FVO was estimated as $11.1 million, which was the lowest the alkali-catalyzed processes. The glycerol credits in Alkali-
among the three processes. FVO and Alkali-WVO processes were $5.49 and $5.19 million,
respectively, which were higher than those in the supercritical
4.2.2. Total manufacturing cost processes due to the high purity of the glycerol.
The results of the total production costs of the three processes The revenue from the sales of biodiesel was $39.6 mil-
are shown in Table 15. The total production cost for Alkali-FVO lion per year in each process on the assumption that all the
was calculated to be $50.9 million, the highest among the pro- biodiesel products were sold in the market. Compared to the
cesses. The lowest total production cost was $25.9 million of total manufacturing costs, the revenue was large enough to
SC-WVO process. Fig. 6 shows the contributions of sub-items make profits in Alkali-WVO and SC-WVO processes. However,
to the total production costs. The oil feed cost was found to be the total manufacturing cost was $6.2 million dollars higher
the major contributor to the total biodiesel production cost in than the earnings from the biodiesel sales in Alkali-FVO pro-
all the processes. As shown in Fig. 6, it comprises 84.0%, 63.9%, cess, indicating that the process was in deficit.
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642 2635

Table 10 – Equipment summary for SC-WVO process.

Pumps P-101 P-102

Type Gear Gear

Efficiency 75% 75%
MOC 316SS 316SS
Flow (kg/h) 5433 5162
Press in (kPa) 101.3 101.3
Press out (kPa) 19000 19000
Driver power (kW) 49.55 39.66

Heat exchangers E-101 E-102 E-103 E-104 E-105

Type Float head Float head Float head Fixed head Float head
Area (m2 ) 24.67 11.53 65.54 53.99 38.04
MOC 316SS 316SS 316SS CS CS
Duty (kW) 992.2 707.3 1460.7 1533.6 514.7
Shell in (◦ C/kPa) 64.76/19000 41.94/19000 385/2533 69.23/101.5 170.8/106.5
Tube in (◦ C/kPa) 287.7/19000 350.0/19000 226.5/19000 35.00/344.98 35.00/344.98

Vessels/Reactors PFR-100 V-100 V-101

Type Continuous pressure vessel Knock-out L-L separator

Orientation Horizontal Vertical Horizontal
MOC 316SS CS CS
T (◦ C) 350 96.18 25.0
P (kPa) 19000 200.0 170.0
Residence time (min) 13.3 10 60
Diameter (m) 0.8 1.22 1.76
Height/length (m) 14.38 3.81 5.28

Tower T-101 T-102

Application Distillation Distillation

# of actual stages 6 10
Type of trays/Packing materials 2 in. Pall rings Valve
MOC CS CS
Diameter (m) 0.765 2.90
Height (m) 6.858 7.62

Table 11 – Investment analysis parameters for the three biodiesel production processes.
Name Value Units

Economic life of project 10 year

Tax rate 40 %/year
Desired rate of return/Interest rate 20 %/year
Salvage value 0 % of initial capital cost
Depreciation method Straight line

Escalation parameters
Project capital escalation 5 %/year
Products escalation 5 %/year
Raw material escalation 3.5 %/year
Operating and maintenance labour escalation 3 %/year
Utilities escalation 3 %/year

Project capital parameters

Working capital percentage 15 % of fixed capital cost

Operating costs parameters

Operating supplies 10 % of maintenance/year
Laboratory charges 16.3 % operating labour/year
Maintenance 6.0 % of fixed capital cost
Plant overhead 50 % of operating labour and maintenance
costs/year
G&A expenses 15 % of subtotal operating costs/year

Facility operation parameters

Facility type Chemical processing facility
Operating mode Continuous processing—24 hours
Length of start-up period 20 weeks
Operating hours per period 8000 hours/year
Process fluids Liquids and gases
2636 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642

Table 12 – The prices of raw materials and products used Table 13 – The prices of utilities used in this study.
in this study.
Utility Price
Name Specification Price ($/tonne) a
Chilled water $1/tonne
Methanol – 268.71a Cooling watera $0.013/tonne
Fresh vegetable oil Canola oil 916b Circulating heating oil $0.4/tonne
Waste vegetable oil – 530c High-pressure steam (4201 kPa)b $16.64/tonne
Medium-pressure steam (1135 kPa)b $13.71/tonne
Glycerol Pharmaceutical 1322.77d
Low-pressure steam (690 kPa)b $12.68/tonne
grade
Vegetable glycerine 860d a
Seider et al. (2004).
b
Sodium hydroxide a
– 200 f Turton (2003).
Sulfuric acida – 60f
Phosphoric acida – 340f
Biodiesel (B100) Qualified to meet EN 990c, g
selling biodiesel product is the same as total manufacturing
14214 and ASTM D
cost (Zhang et al., 2003). Since Table 16 shows before-tax val-
6751
ues, taxes are not applied to the break-even prices.
a
McGinn (2009). The net present values (NPVs) were positive except for
b
USDA (2010a). Alkali-FVO process. SC-WVO process showed much higher
c
USDA (2010b). NPV than the alkali-catalyzed processes due to the lower total
d
Lefebvre (2009).
f
manufacturing costs. For Alkali-FVO process, the margin from
West et al. (2008).
g biodiesel and glycerol sales was not high enough to calcu-
Density of 870.5 kg/m3 was assumed.
late the discounted payback period (DPP) and discounted cash
flow rate of return (DCFROR). The DCFROR were 22.4% and
4.2.3. Profitability analysis 49.6% for Alkali-WVO and SC-WVO processes, respectively.
Table 16 shows the profitability measures for the three Since these values were higher than the minimum accept-
processes including break-even prices of biodiesel. The break- able rate of return (MARR) which was set at 20% and the NPVs
even price was defined as the price at which the revenue from were greater than zero, the two processes were considered

Table 14 – Total capital investment for biodiesel production processes at the capacity of 40,000 tonne/yr ($ millions).
Type Description Alkali-FVO Alkali-WVO SC-WVO

Direct items
Reactor Esterification – 0.255 –
Transesterification 0.252 0.249 0.726
Neutralization 0.080 0.080 –

Column Methanol recoverya 0.093 0.256b /0.099 0.108

Biodiesel purificationa 0.186 0.182 0.227
Glycerol purificationa 0.065 0.047 –
Water washing 0.034 0.034 –
Glycerol washing – 0.073 –

Other Pumps 0.017 0.031 0.052

Heat exchangers 0.020 0.059 0.557
L-L separator 0.025 0.027 0.026
Flash evaporator – – 0.021

Total purchased equipment 0.772 1.39 1.72

Equipment setting 0.016 0.023 0.023
Piping 0.690 1.12 1.77
Civil 0.133 0.187 0.126
Steel 0.054 0.087 0.037
Instrumentation 1.21 1.68 0.991
Electrical 0.497 0.529 0.489
Insulation 0.195 0.283 0.246
Paint 0.028 0.035 0.021
Total direct cost 3.594 5.336 5.425
Othersc 4.050 5.454 3.703
G&A Overheads 0.150 0.219 0.209
Contract Fee 0.402 0.526 0.412
Contingencies 1.48 2.08 3.17
Fixed capital cost 9.67 13.6 12.9
Working capital 1.45 2.04 1.94
Total capital investment 11.1 15.6 14.8

a
Includes the costs for a distillation column, condenser, distillate accumulator, reflux pump and reboiler.
b
Indicates a pre-treatment process.
c
This item is the total of the following costs: design, engineering, and procurement costs, material charges (freight and taxes) and construction
field indirect costs (fringe benefits, burdens, consumables/small tolls, insurance, equipment rental, field services, field office construction
supervision, and plant start-up).
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642 2637

Table 15 – Total manufacturing cost, glycerol credit and revenues from biodiesel sales in the three processes at the
capacity of 40,000 tonne of biodiesel production/yr ($ millions).
Process Alkali-FVO Alkali-WVO SC-WVO

Direct manufacturing cost

Oil feed 38.46 22.36 21.88
Methanol 1.215 1.356 1.24
Catalysts 0.201 0.215 –
Water washing 0.000 0.000 –
Glycerol washing – 3.296 –
Steam (4201 kPa) – – –
Steam (1135 kPa) 0.006 0.007 0.168
Steam (690 kPa) 0.132 0.563 –
Circulating heating oil 0.407 0.400 0.616
Cooling water 0.019 0.019 0.030
Chilled water 0.234 1.745 –
Electricity 0.066 0.099 0.089
Operating labour 0.800 0.800 0.600
Supervision 0.280 0.280 0.280
Maintenance 0.580 0.817 0.775
Operating supplies 0.058 0.082 0.078
Laboratory charges 0.143 0.176 0.143
Subtotal 42.40 32.22 25.90

Indirect manufacturing cost

Overhead 0.730 0.948 0.828
Depreciation 0.822 1.350 1.096
Property Taxes 0.224 0.315 0.299
Insurance 0.097 0.136 0.129
Subtotal 1.873 2.749 2.352
General and Administrative costs 6.641 5.245 4.238
Total production cost 50.9 40.2 32.49
Glycerol credit 5.49 5.19 3.45
Total manufacturing cost 45.8 35.0 29.0
Revenue from biodiesel 39.6 39.6 39.6

economically profitable. Overall, SC-WVO was evaluated as the absolute value of the percentage of the oil price in Fig. 8 is
the most favourable process, followed by Alkali-WVO. Alkali- higher in Alkali-FVO than the other processes due to the high
FVO was considered to be an unacceptable project. price of the fresh vegetable oil. Alkali-FVO process was, how-
ever, more sensitive to the oil price than any other processes
4.3. Sensitivity analysis on net present value even when the same absolute values of the changes in oil price
were considered. The slopes of the trend lines of each process
Sensitivity analyses for the three biodiesel production pro- were evaluated and shown in Fig. 7.
cesses were conducted to investigate sensitivity of the NPVs Fig. 8 shows the effect of changes in glycerol price on the
to changes in a variety of parameters. Five parameters were NPV of each process. The results indicated that an increase
investigated: price of oil feed; glycerol credit; biodiesel selling of $0.89 million in NPV of SC-WVO could be caused by an
price; interest rate; and the life of a plant. The prices of oil feed, increase of $0.086/kg in the price of crude glycerol. An increase
glycerol and biodiesel were varied by ±10, 20, 30 and 40% from of $0.132/kg in the price of pharmaceutical grade glycerol
the original values shown in Table 13. The interest rate varied resulted in an increase of $2.1 and $1.31 million in the NPV
from 10% to 40%. The expected life of the plant was prolonged of Alkali-FVO and Alkali-WVO, respectively. For Alkali-WVO
to 15 and 20 years. process, the NPV turned to negative value as the glycerol
The sensitivity analysis results of the oil prices are shown price was decreased by 15%. Overall, the influence of glyc-
in Fig. 7. The NPVs of the three processes were significantly erol credit is much less than the influence of the prices of oil
changed by the change in the prices of oil feed. The results feed.
showed that every 10% increase in the oil price approximately The sensitivity analysis results of biodiesel selling price are
decreased the NPV by $15, $6.8, and $6.4 million in Alkali-FVO, shown in Fig. 9. The ±10% changes in the price of biodiesel
Alkali-WVO, and SC-WVO, respectively. It should be noted that ($0.099/kg) changed the NPVs of Alkali-FVO, Alkali-WVO, and

Table 16 – The profitability measures and the break-even price of biodiesel.

Alkali-FVO Alkali-WVO SC-WVO

Discounted payback period (years) – 9.2 4.2

Net present value ($ millions) −26.697 1.901 21.086
Discounted cash flow rate of return (%) – 22.4 49.6
Break-even price of biodiesel ($/kg) 1.15 0.875 0.726
Break-even price of biodiesel ($/l)a 1.00 0.762 0.632

a
Density of 870.5 kg/m3 was used.
2638 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642

Fig. 6 – The proportions of subitems in the total manufacturing cost of biodiesel of the three processes.

Fig. 7 – Sensitivities of the NPVs of the three processes to the price of oil feeds.

Fig. 8 – Sensitivities of the NPVs of the three processes to the price of glycerol by-product.
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642 2639

Fig. 9 – Sensitivities of the NPVs of the three processes to the price of biodiesel product.

SC-WVO by ±$13.7, $11.1, and $10.3 million, respectively. The research findings on the effect of multiple parameters on the
results also indicate that the NPV of Alkali-WVO and SC-WVO profitability of biodiesel production processes. Therefore, in
processes are more strongly influenced by biodiesel selling order to investigate the effects of simultaneous changes in the
prices than the oil feed prices. For Alkali-FVO process, the parameters statistical analysis using design of experiments
effect of the oil price was more significant than the biodiesel was performed.
price, playing the most important role in the profitability of The method selected was Response Surface Method (RSM)
Alkali-FVO. coupled with Central Composite Design (CCD) using JMP 8
The sensitivity analysis of the interest rate is shown in software (SAS Institute Inc.). Table 17 shows the four param-
Fig. 10. The results indicated that the NPVs of the three eters and their five different levels (−˛, −1, 0, +1, +˛) used in
processes were very sensitive to the interest rate used for the analysis. After experimental tables were made for each of
discounting cash flows. Since cash flows are discounted to the processes, the NPV was calculated using IPE accordingly.
time 0, the cash flows closer to the end of the project are Table 18 shows the results of the Analysis of Variance (ANOVA)
heavily discounted. Therefore, as the interest rate increases, of Alkali-FVO process. The parameters with Prob >F (p-value)
the revenues from biodiesel sales near the end of the project less than 0.05 were considered significant and marked with an
are heavily discounted, whereas the capital investment at asterisk. Coefficients of each term in a regression equation to
time 0 remains constant. As a result, the NPVs of Alkali- calculate NPV are shown as “Estimate”.
WVO, and SC-WVO decreased as the interest rate increases. The predicted formula for NPV was derived from the
On the other hand, the deficit financing in Alkali-FVO over linear combination of the parameters and the values of
the life of the project became heavily discounted as the inter- their corresponding estimates after eliminating insignificant
est rate increases and it resulted in the increases in the parameters, Eq. (1):
NPVs.
In reality, however, most of the parameters simultaneously NPV ($ million) = −26.697 + 57.9212A − 63.9471B + 8.4131C
and continuously change over time in an irregular way and the
+ 7.3551D+12.7501AB − 24.8585AD + 26.084BD − 5.6084A2
actual values of these parameters are occasionally unknown
at the time of evaluation. Nevertheless, there are not sufficient −6.2336B2 (1)

where A = (biodiesel price ($/kg) − 0.99)/0.396, B = (FVO price

($/kg) − 0.916)/0.3664, C = (glycerol price ($/kg) − 1.323)/0.5292,
and D = (interest rate (%) − 20)/10.
A, B, C and D are expressed in the form of (Xi − Xi,0 )/Xi
where Xi,0 is the value at level 0 and Xi is a difference value
of adjacent levels in Table 17.
Fig. 11 shows the comparison of actual NPV values and
those calculated by Eq. (1) for Alkali-FVO process. The solid
and dotted lines represent the line of perfect fit and the 0.05
significance curve, respectively. The calculated R2 was 0.99.
Therefore, it is expected that the formula can be used as
relatively simple means to predict profitability of Alkali-FVO
process from any of the combinations of the oil feed price,
glycerol price, biodiesel selling price and interest rate within
the ranges in Table 17.
Fig. 10 – Sensitivities of the NPVs of the three processes to The NPV correlations of Alkali-WVO and SC-WVO pro-
interest rates. cesses were developed in the same manner. For Alkali-WVO
2640 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642

Table 17 – The four parameters in central composite design to investigate NPVs of Alkali-FVO, Alkali-WVO, and SC-WVO
processes.
Variables Specification Unit Level

−2 (−˛) −1 0 +1 +2 (+˛)

Oil feed price FVO $/kg 0.1832 0.5496 0.916 1.2824 1.6488
WVO 0.106 0.318 0.53 0.742 0.954

Biodiesel price – $/kg 0.198 0.594 0.99 1.386 1.782

Glycerol price Pharmaceutical $/kg 0.2646 0.7938 1.323 1.8522 2.3814

Crude 0.172 0.516 0.86 1.204 1.548

Interest rate – % 0 10 20 30 40

Table 18 – Analysis of variance and effect tests for each parameter of Alkali-FVO process.
Analysis of variance

Source Degree of Freedom Sum of Squares Mean Square F ratio Prob >F

Model 14 207532.33 14823.7 233.3583 <0.0001

Effect tests

Term Estimate Sum of squares F ratio Prob >F

Biodiesel price 57.9212 80516.909 1267.514 <0.0001*

FVO price −63.9471 98141.507 1544.964 <0.0001*
Glycerol price 8.4131 1698.719 26.7416 <0.0001*
Interest rate 7.3551 1298.334 20.4386 0.0002*
Biodiesel price × FVO price 12.7501 2601.051 40.9463 <0.0001*
Biodiesel price × Glycerol price −1.0057 16.185 0.2548 0.6190
FVO price × Glycerol price 1.0315 17.024 0.2680 0.6101
Biodiesel price × Interest rate −24.8585 9887.120 155.6451 <0.0001*
FVO price × Interest rate 26.0840 10886.001 171.3697 <0.0001*
Glycerol price × Interest rate −3.5249 198.796 3.1295 0.0914
Biodiesel price × Biodiesel price −5.6084 1006.531 15.8450 0.0007*
FVO price × FVO price −6.2336 1243.467 19.5749 0.0002*
Glycerol price × Glycerol price −0.6793 14.765 0.2324 0.6347
Interest rate × Interest rate −0.4300 5.917 0.0932 0.7632

process:
For SC-WVO process:
NPV ($ million) = 1.6884 + 50.8435A − 34.1184B + 6.3613C
−8.8331D + 7.3361AB − 21.776AD + 14.3841BD − 4.7346A2 NPV ($ million) = 19.587 + 49.4227A − 31.3785B + 4.2480C′
−2.4894B2 + 4.1646D2 (2) −17.4959D − 20.7246AD + 6.0322AB + 13.0959BD − 4.3212A2

+ 6.5714D2 (3)

where A = (biodiesel price ($/kg) − 0.99)/0.396, B = (WVO

price ($/kg) − 0.53)/0.212, C = (pharmaceutical-grade glyc-
erol price ($/kg) − 1.323)/0.5292, C’ = (crude glycerol price
($/kg) − 0.86)/0.344, and D = (interest rate (%) − 20)/10.

5. Conclusion

The normal boiling point of triolein was determined as 412.8 ◦ C

through the thermogravimetric analysis. The HYSYS simula-
tion results showed significant improvements in the accuracy
of triolein properties when TGA boiling point was applied. All
of the three processes were feasible for producing a high qual-
ity biodiesel product which meets EN 14214 standard. The total
energy consumption of the supercritical process was found to
be comparable to that of a homogeneous alkali-catalyzed pro-
cess using fresh vegetable oil as the raw material by utilizing
a flash evaporator to primarily separate methanol from the
reactor effluent stream.
Fig. 11 – A comparative plot of NPV values calculated by the The economic assessment of the three types of
prediction formula to the actual values. biodiesel production plants (40,000 tonne/year): Alkali-FVO,
 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642 2641

Alkali-WVO, and SC-WVO, was performed, based on the pro- Han, B., Peng, D., 1993. A group-contribution correlation for
cess simulation models developed in HYSYS with TGA triolein predicting the acentric factors of organic compounds. Can. J.
as a model compound of vegetable oils. The assessment was Chem. Eng. 71, 332–334.
Hawash, S., Kamal, N., Zaher, F., Kenawi, O., Diwani, G.El., 2009.
done for the first time using Aspen In-Plant Cost Estimator.
Biodiesel fuel from Jatropha oil via non-catalytic supercritical
The results showed that the most significant variable affect- methanol transesterification. Fuel 88, 579–582.
ing the production cost of biodiesel was the oil feed cost, He, H., Wang, T., Zhu, S., 2007. Continuous production of
comprising 64–84% of the biodiesel manufacturing costs. The biodiesel fuel from vegetable oil using supercritical methanol
estimated DPP, NPV and DCFROR indicated that SC-WVO process. Fuel 86, 442–447.
was the most financially promising process, followed by Kasteren, J.M.N.V., Nisworo, A.P., 2007. A process model to
Alkali-WVO, while Alkali-FVO was found to be non-profitable. estimate the cost of industrial scale biodiesel production from
waste cooking oil by supercritical transesterification. Resour.
Among the parameters investigated in the sensitivity
Conserv. Recycling 50, 442–458.
analyses, biodiesel selling price, prices of feedstock oil and Kusdiana, D., Saka, S., 2001. Kinetics of transesterification in
by-product glycerol and interest rate were found to be the rapeseed oil to biodiesel fuel as treated in supercritical
most significant variables affecting the economic viability of methanol. Fuel 80, 693–698.
the three processes. The four parameters were further statis- Kusdiana, D., Saka, S., 2004. Effects of water on biodiesel fuel
tically analyzed based on the design of experiment approach production by supercritical methanol treatment. Bioresour.
Technol. 91, 289–295.
using Response Surface Method coupled with Central Com-
Lefebvre, B., 2009. Glycerine Price Sample Report (US Gulf), ICIS
posite Design. From the results, prediction formulas of the
Pricing. Available at:
NPV were derived for each of the three processes. The formu- http://www.icispricing.com/il shared/Samples/SubPage170.asp
las can be used for various combinations of the four significant (Accessed April 19, 2010).
parameter values to predict economic feasibility. Lim, Y., Lee, H., Lee, Y., Han, C., 2009. Design and economic
analysis of the process for biodiesel fuel production from
Acknowledgments transesterificated rapeseed oil using supercritical methanol.
Ind. Eng. Chem. Res. 48, 5370–5378.
Luyben, W.L., 2002. Plantwide dynamic simulators in chemical
The authors acknowledge the financial support of the Natural processing and control. CRE Press, New York.
Sciences and Engineering Research Council of Canada. Madras, G., Kolluru, C., Kumar, R., 2004. Synthesis of biodiesel in
supercritical fluids. Fuel 83, 2029–2033.
Marulanda, V.F., Anitescu, G., Tavlarides, L.L., 2010. Biodiesel
References fuels through a continuous flow process of chicken fat
supercritical transesterification. Energy Fuels 24, 253–260.
Ambrose, D., 1978. Correlation and Estimation of Vapor-liquid McGinn, S., 2009. Methanol Price Sample Report (US Gulf), ICIS
Critical Properties: I. Critical Temperatures of Organic Pricing. Available at:
Compounds. National Physical Laboratory Report (U.K.), Div. http://www.icispricing.com/il shared/Samples/SubPage135.asp
Chem., 92. (Accessed April 19, 2010).
Bunyakiat, K., Makmee, S., Sawangkeaw, R., Ngamprasertsith, S., Perry, E.S., Weber, W.H., Daubert, B.F., 1949. Vapor pressures of
2006. Continuous production of biodiesel via phlegmatic liquids, I. simple and mixed triglycerides. J. Am.
transesterification from vegetable oils in supercritical Chem. Soc. 71, 3720–3726.
methanol. Energy Fuels 20, 812–817. Przybylski. Canola Oil: Physical and Chemical Properties, 2008.
Ceriani, R., Meirelles, A.J.A., 2004. Predicting vapor-liquid Available at:
equilibria of fatty systems. Fluid Phase Equilib. 215, 227–236. http://www.canolacouncil.org/uploads/technical/chemical1-
Chopey, N.P., 2004. Handbook of Chemical Engineering 6/chemical1-6 1.html (Accessed Sept 18, 2010).
Calculations. McGraw-Hill. Pinnarat, T., Savage, P.E., 2008. Assessment of noncatalytic
Constantinou, L., Gani, R., 1994. New group contribution method biodiesel synthesis using supercritical reaction conditions.
for estimating properties of pure compounds. AIChE J. 40, Ind. Eng. Chem. Res. 47, 6801–6808.
1697–1710. Reid, R.C., Prausnitz, J.M., Poling, B.E., 1987. The Properties of
Constantinou, L., Gani, R., O’Connell, J.P., 1995. Estimation of the Gases and Liquids, fourth ed. McGraw Hill, New York.
acentric factor and the liquid molar volume at 298 K using a Santana, G.C.S., Martins, P.F., Silva, N., Batistella, C.B., Filho, R.M.,
new group contribution method. Fluid Phase Equilib. 103, Maciel, M.R.W., 2010. Simulation and cost estimate for
11–22. biodiesel production using castor oil. Chem. Eng. Res. Des. 88,
Dohrn, R., Brunner, G., 1991. Correlations of pure-component 626–632.
parameters of the Peng-Robinson equation of state. In: Seider, W.D., Seader, D., Lewin, D.R., 2004. Product and Process
Proceedings of Second International Symposium on Design Principles: Synthesis, Analysis and Evaluation, second
Supercritical Fluids, Boston, pp. 471–478. ed. Wiley, New Jersey.
Dohrn, R., Brunner, G., 1994. An estimation method to calculate Tang, Z., Du, Z., Min, E., Gao, L., Jiang, T., Han, B., 2006. Phase
Tb, Tc, Pc and ␻ from the liquid molar volume and the vapor equilibria of methanol-triolein system at elevated
pressure. In: Proceedings of Third International Symposium temperature and pressure. Fluid Phase Equilib. 239, 8–11.
on Supercritical Fluids, Strasbourg, France, pp. 241– Turton, R., 2003. Analysis, Synthesis and Design of Chemical
248. Processes, second ed. Prentice Hall, New Jersey.
Glisic, S., Montoya, O., Orlovic, A., Skala, D., 2007. Vapor–liquid USDA Foreign Agricultural Service, 2010a. Oilseeds: World
equilibria of triglycerides–methanol mixtures and their Markets and Trade. Available at:
influence on the biodiesel synthesis under supercritical http://www.fas.usda.gov/oilseeds/circular/2010/March/
con-ditions of methanol. J. Serb. Chem. Soc. 72, 13–27. oilseedsfull03-10.pdf (Accessed April 19, 2010).
Glisic, S., Lukic, I., Skala, D., 2009. Biodiesel synthesis at high USDA Agricultural Marketing Service, 2010b. National Weekly Ag
pressure and temperature: analysis of energy consumption on Energy Roundup. Available at:
industrial scale. Bioresour. Technol. 100, 6347–6354. http://www.ams.usda.gov/mnreports/lswagenergy.pdf
Goodrum, J.W., Geller, D.P., 2002. Rapid thermogravimetric (Accessed April 19, 2010).
measurements of boiling points and vapor pressure of Varma, M.N., Madras, G., 2007. Synthesis of biodiesel from castor
saturated medium-and long-chain triglycerides. Bioresour. oil and linseed oil in supercritical fluids. Ind. Eng. Chem. Res.
Technol. 84, 75–80. 46, 1–6.
2642 chemical engineering research and design 8 9 ( 2 0 1 1 ) 2626–2642

Weber, W., Petkov, S., Brunner, G., 1999. Vapour-liquid-equilibria You, Y., Shie, J., Chang, C., Huang, S., Pai, C., Yu, Y., Chang, C.,
and calculations using the Redlich-Kwong-Aspen-equation of 2008. Economic cost analysis of biodiesel production: case in
state for tristearin, tripalmitin, and triolein in CO2 and soybean oil. Energy Fuels 22, 182–189.
propane. Fluid Phase Equilib. 158–160, 695–706. Zhang, Y., Dubè, M.A., McLean, D.D., Kates, M., 2003. Biodiesel
West, A.H., Posarac, D., Ellis, N., 2008. Assessment of four production from waste cooking oil: 1, Process design and
biodiesel production processes using HYSYS. Plant. Bioresour. technological assessment. Bioresour. Technol. 89,
Technol. 99, 6587–6601. 1–16.