Bioresource Technology xxx (2013) xxxxxx

Contents lists available at SciVerse ScienceDirect

Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech

Biodiesel production with continuous supercritical process:

Non-catalytic transesterication and esterication with or without
carbon dioxide
Yu-Ting Tsai, Ho-mu Lin, Ming-Jer Lee
Department of Chemical Engineering, National Taiwan University of Science and Technology, 43 Keelung Road, Section 4, Taipei 106-07, Taiwan

h i g h l i g h t s g r a p h i c a l a b s t r a c t

" Biodiesel production with

supercritical uid technology has
been studied.
" Kinetic behavior of
transesterication and esterication
was investigated.
" A continuous tubular reactor was
employed in the kinetics study.
" Waste cooking oil was also used as a
feedstock.
" A power law model correlated well
the kinetic data.

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

Article history: The non-catalytic transesterication of rened sunower oil with supercritical methanol, in the presence
Available online xxxx of carbon dioxide, was conducted in a tubular reactor at temperatures from 553.2 to 593.2 K and pres-
sures up to 25.0 MPa. The FAME yield can be achieved up to about 0.70 at 593.2 K and 10.0 MPa in
Keywords: 23 min with methanol:oil of 25:1 in molar ratio. The effect of adding CO2 on the FAME yield is insignif-
Biodiesel icant. The kinetic behavior of the non-catalytic esterication and transesterication of oleic acid or waste
Continuous supercritical process cooking oil (WCO) with supercritical methanol was also investigated. By using the supercritical process,
Transesterication
the presence of free fatty acid (FFA) in WCO gives positive contribution to FAME production. The FAME
Esterication
yield of 0.90 from WCO can be achieved in 13 min at 573.2 K. The kinetic data of supercritical transeste-
rication and esterifaication were correlated well with a power-law model.
2012 Elsevier Ltd. All rights reserved.

1. Introduction et al., 2010), or lipase-catalyst (Shieh et al., 2003) is generally re-

quired in BD production processes. Homogeneous alkaline catalyst
Biodiesel (BD) has been recognized as one of potential alterna- is commonly used in the worldwide commercial BD plants
tive renewable fuels in the future. In recent years, various articles (Mittelbach and Remschmidt, 2006). This process, however, needs
extensively reviewed the progress made in the developments of BD high quality feedstock and catalyst removal from the product
production technology using a wide variety of raw materials, mixtures, resulting that the complicated purication operations
including edible oils and fats (Ma and Hanna, 1999; Yusuf et al., are required. Using heterogeneous catalysts may simplify the
2011), non-edible oils (Juan et al., 2011; Wang et al., 2012), micro- product recovery process (Semwal et al., 2011). Unfortunately,
algae (Cheng and Timilsina, 2011), etc. To promote the transesteri- the performance of most acidic heterogeneous catalysts is still
cation of oils with methanol, basic (Keera et al., 2011; Sharma unacceptable due to slow reaction rates and inducing unfavorable
et al., 2011), acidic (Lotero et al., 2005), acidicbasic (Macario side reactions. The reaction rates become faster using basic heter-
ogeneous catalysts, but their activities may be degraded in the
Corresponding author. Tel.: +886 2 2737 6626; fax: +886 2 2737 6644. presence of water. While some promising results were found by
E-mail addresses: [email protected], [email protected] (M.-J. Lee). using lipase-catalyst, the reaction rates are rather slow.

0960-8524/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.biortech.2012.12.157

Please cite this article in press as: Tsai, Y.-T., et al. Biodiesel production with continuous supercritical process: Non-catalytic transesterication and ester-
ication with or without carbon dioxide. Bioresour. Technol. (2013), http://dx.doi.org/10.1016/j.biortech.2012.12.157
2 Y.-T. Tsai et al. / Bioresource Technology xxx (2013) xxxxxx

Alternatively, non-catalytic supercritical methanol process has Waste cooking oil (WCO) is one of cheap raw materials for
been considered as a potential method for BD production without producing BD. The non-catalytic supercritical methanol process,
suffering the above mentioned shortcomings (Saka and Kusdiana, which can tolerate certain amounts of impurities (mainly, water
2001; Kusdiana and Saka, 2004; Demirbas, 2006; Marulanda and FFA) in the feedstock, could be a favorable method for convert-
et al., 2010; de Boer and Bahri, 2011; Tan and Lee, 2011). This ing WCO into BD. In this BD production process, the esterication
supercritical process allows feedstock containing certain amount of FFA with methanol takes place simultaneously with the transe-
of water and FFA. High reaction rate and without using any catalyst sterication of WCO with methanol. As a consequence, the kinetic
are also its advantages over the traditional BD production method. models for representing the esterication of FFA with methanol
Saka and Kusdiana (2001) rst applied supercritical technique to and the transesterication of oils with methanol are needed in
BD production. It was found that about 0.95 of FAME yield can the process development of BD production. The former model is
be achieved in 2 min at 623 K and 45 MPa with a methanol to oil also essential for development of the two-step supercritical meth-
ratio of 42:1. Operating at such high temperatures, the quality of anol process (Minami and Saka, 2006). Up to date, the kinetic stud-
product may be deteriorated due to thermal decomposition (Song ies on the esterication of FAA with supercritical methanol are still
et al., 2008). Imahara et al. (2008) investigated the decomposition limited. Minami and Saka (2006) presented graphical results for
behavior with FT-IR analysis and found that FAMEs were stable at the esterication. Alenezi et al. (2010) investigated the esterica-
543 K, but partly decomposed at 623 K. To produce high quality BD tion of FFA (containing oleic acid 88 wt.%) with supercritical meth-
and reduce the energy consumption, lowering the reaction temper- anol in a batch reactor at temperature range from (523 to 593) K
ature and pressure are key factors to improve the BD production and 10 MPa with methanol to FFA ratios from 0.7:1 to 7:1.
with supercritical technology. In the present study, experimental runs of non-catalytic esteri-
Adding proper co-solvent may lower the critical points of the cation of oleic acid with supercritical methanol were conducted
reacting mixtures, which may result in decrease of required tem- in a tubular reactor at (493.2533.2) K, 10.0 MPa, and methanol
perature and pressure in the reactor. Han et al. (2005) used CO2 to oleic acid ratios from 2:1 to 5:1. The kinetics of WCO with meth-
as co-solvent and reported that about 0.98 of FAME yield was ob- anol was also investigated at 573.2 K and 10.0 MPa by using the
tained from a batch reactor in 10 min at 553 K, 14.3 MPa, methanol same continuous reactor. Furthermore, the kinetic data of super-
to oil molar ratio of 24:1, and CO2 to methanol molar ratio of 1:10. critical transesterication and esterication were correlated with
However, Imahara et al. (2009) claimed that the addition of CO2 a power-law model, respectively, and the kinetic parameters of
decrease FAME yield by using a continuous tubular reactor. The both reaction systems were determined.
highest FAME yield is no greater than 0.80 at 593 K and 20 MPa.
Similar co-solvent effect was observed from the transesterication
of soybean oil and ethanol through a tubular reactor with inner 2. Methods
diameter 3.2 mm (Bertoldi et al., 2009). Using a microtube reactor
with inner diameter 0.76 mm, Silva et al. (2010) reported the addi- 2.1. Materials
tion of CO2 did not signicantly affect the ethyl esters yields but
Trentin et al. (2011) found that the yield increased with increasing Rened sunower oil (Quaker) was purchased from Hypermall
the co-solvent addition to the reacting system. In the present (Taiwan), methanol (dried, <0.005% H2O), acetone (99.9%+), oleic
study, we attempt to clarify the co-solvent (CO2) effect on the acid (technical grade, 90%), eicosane (99%), and heptadecanoic acid
supercritical transesterication through a continuous operation (98%+) from SigmaAldrich, FAME standard (FAME Mix GLC-10,
mode. The experiments of non-catalytic transesterication of re- methyl palmitate 20 wt.%, methyl stearate 20 wt.%, methyl oleate
ned sunower oil with supercritical methanol, in the presence 20 wt.%, methyl linoleate 20 wt.%, methyl linolenate 20 wt.%) from
of carbon dioxide, were thus conducted in a tubular reactor at tem- Supelco (USA). Waste cooking oil was supplied by Chant Oil Co. Ltd.
peratures from 553.2 to 593.2 K and pressures from 10.0 to (Taiwan). Its composition is listed in Table 1. Carbon dioxide
25.0 MPa. (99.5%+) was supplied from Liu-Hsiang Co. (Taiwan). Water used
Another possible way to reduce the required reaction tempera- for preparing aqueous solutions was obtained from NANO pure-Ul-
ture and pressure was proposed by Minami and Saka (2006). They tra pure water system that was distilled and deionized with resis-
developed a two-step supercritical methanol process. It included tance of 18.3 MX. All of the chemicals were used without further
the hydrolysis of oil to producing fatty acids and then followed purication.
the esterication of fatty acids with supercritical methanol to syn-
thesis of methyl esters. Under milder conditions, at 563 K and 2.2. Experimental conditions and procedure
20 MPa, 0.90 FAME yield can be obtained in 15 min via esterica-
tion of oleic acid with supercritical methanol. Moreover, less A continuous supercritical BD production system was installed.
excess methanol is necessary in this two-step process, resulting Its schematic diagram is shown in Fig. 1. This system was equipped
in reducing the energy consumption for recovery of unreacted with three feeding pumps. One syringe pump (2, model 500D, Isco,
methanol. Dippolito et al. (2007) simulated a supercritical BD pro- USA; uncertainty of ow rate = 0.5%, maximum ow rate =
duction process with heat recovery by using two successive reac-
tors together with intermediate removal of glycerol. The
simulation results showed that the operating pressure can be de-
creased to (1011) MPa, methanol/oil ratio to 10 and temperature Table 1
Composition and properties of the waste cooking oil (WCO).
to (548563) K. Under these operating conditions, the pumping
energy can be reduced by 23%. It appears that heat integration, de- Triglyceride/wt.% 92.27
crease of required reaction pressure and excess methanol, or using Diglyceride/wt.% 3.03
Monoglyceride/wt.% 0.05
the hydrolysis-esterication two-step process may substantially Free fatty acid/wt.% 4.44 16:0 18:0 18:1 18:2 18:3
improve energy consumption in the supercritical BD production 25% 17% 34% 23% 0%
processes. Water/wt.% 0.080
Since the raw materials cost accounts for about (7080)% of to- Acid value (mg KOH g oil1) 5.014
Density (at 298 K, 0.1 MPa) 0.914
tal cost of BD production (Yusuf et al., 2011), using low-grade feed-
Average molecular weight 793.940
stock should be one of key factors to reduce the production cost.

Please cite this article in press as: Tsai, Y.-T., et al. Biodiesel production with continuous supercritical process: Non-catalytic transesterication and ester-
ication with or without carbon dioxide. Bioresour. Technol. (2013), http://dx.doi.org/10.1016/j.biortech.2012.12.157
 Y.-T. Tsai et al. / Bioresource Technology xxx (2013) xxxxxx 3

1. CO2 cylinder 6. Coil preheaters 11. Needle valve

2. Syringe pump 7. Thermostatic air bath 12. Pressure regulator
3. HPLC pump 8. Reactor 13. Pressure release cylinder
4. Low temperature 9. Temperature 14. Sample collector
circulation indicator
5. Feeding bottles 10. Pressure indicator
Fig. 1. The schematic diagram of the continuous reaction system.

204 cm3 min1) was utilized to charge carbon dioxide, if necessary. 2.3. FAME analysis
Another syringe pump (2, model 260D, Isco, USA; uncertainty of
ow rate = 0.5%; maximum ow rate = 107 cm3 min1) delivered The composition of the product samples was analyzed by a gas
oils into reactor. Meanwhile, methanol is fed by an HPLC pump chromatography (Varian CP 3800, USA) equipped with a ame-ion-
(3, model PU-2080, Jasco, Japan; uncertainty of ow rate = 2%; ization detector and using high purity helium as carrier gas. A cap-
maximum ow rate = 10 cm3 min1). Each feed stream was illary column (CP9080, Select Biodiesel for FAME,
preheated in a coil (6) before entering a tubular reactor (8, stainless 30 m  0.32 mm ID, Varian, USA) was used for the composition
steel tube packed with stainless steel tailings, internal free volume analysis of the samples taken from the products of transesterica-
about 50 cm3, inner diameter of the tube 3.86 mm). The preheaters tion of rened sunower oil with methanol and the esterication
and the reactor were placed in an air bath (7). Temperature of bath of oleic acid with methanol. The ow rate of helium was xed at
was measured by a digital thermometer (9) and controlled to 1 cm3 min1. The temperatures of injector and detector were set
within 0.5 K. Pressure in the reactor was measured by a pressure at 573.2 K. The oven temperature was starting from 435.2 K for
transducer (10, PDCR 330, maximum pressure 35 MPa, Druck, UK) 36 min, then increasing to 463.2 K with the rising rate of
with a digital indicator (DPI 280, Druck, UK). The uncertainty of 20 K min1. After holding at 463.2 K for 4 min, the oven tempera-
pressure reading is 0.1%. A needle valve (11) and a pressure ture increased to 513.2 K within 2 min. The analysis was
regulator (12) were adjusted manually to manipulate the reaction terminated after holding at 513.2 K for 15 min. The split ratio
pressure to within 0.1 MPa. The product was cooled in a pressure was set to be 100 for 5 min and then reduced to 20.
release cylinder (13) and collected in a sample bottle (14). Another column (Supelcowax 10, 30 m  0.53 mm ID 
Product samples were collected after the operation reaching a 1.0 lm, Supelco, USA) was used for the composition analysis of
steady-state. This apparatus was used in all the kinetic studies of the samples taken from the product of the WCO with methanol.
this work. The injector and detector temperatures were xed at 553.2 K.
Each collected sample was washed with hot water at about The oven temperature was maintained at 483.2 K for 7 min, then
363.2 K for three times to remove glycerol and unreacted metha- increasing to 493.2 K within 2 min. After holding at 493.2 K for
nol. After centrifugation, the upper organic (FAME-rich) phase 10 min, the temperature was elevated to 513.2 K with rising rate
was separated from the lower aqueous phase for composition anal- of 3 K min1 and then maintained at 513.2 K for 15 min. Finally, in-
ysis. A given amount of heptadecanoic acid/acetone solution was creased the temperature up to 533.2 K within 1 min and holding at
used as internal standard for composition determination with a this temperature for 9 min before ending the analysis. Initially, the
gas chromatography. The experiments were operated over a wide split ratio was set to be 50 for 5 min and then reduced to 20.
range of ow rates, corresponding to different average residence At least four replicated samples were taken at each experimental
times in the reactor. The volumetric ow rate of each feed stream condition. The reproducibility was found to within 2%. The mass
was periodically calibrated with pure water. The uncertainty of ow rate of FAME in the product stream (WFAME,out) was calculated
reported residence time was estimated to be 1%. with the aid of the calibration curves, which were prepared by using

Please cite this article in press as: Tsai, Y.-T., et al. Biodiesel production with continuous supercritical process: Non-catalytic transesterication and ester-
ication with or without carbon dioxide. Bioresour. Technol. (2013), http://dx.doi.org/10.1016/j.biortech.2012.12.157
4 Y.-T. Tsai et al. / Bioresource Technology xxx (2013) xxxxxx

FAME standard (FAME Mix GLC-10, Supelco, USA) and heptadeca- conducted at 553.2 K, methanol to oil ratio (hB0) of 42, CO2 to meth-
noic acid as internal standard. The FAME yield was calculated from anol ratio (hE0) of 0.1, the residence time (s) of 18.6 min and pres-
FAME yield W FAME;out =W TG0 1 sures over 10.025.0 MPa. The results show that the FAME yields
distribute in the range of 0.280.32 as pressure changes over the
where WTG0 is the mass ow rate of triglyceride (oil) in the feed whole pressure range. It appears that the FAME yield is insensitive
stream. The uncertainty of the reported FAME yield was estimated to pressure over the experimental condition range. As a conse-
to be 2.3%. quence, the reaction pressure is xed at 10.0 MPa for the subse-
quent experiments.
3. Experimental results and discussion
3.1.2. Feed composition effect
3.1. Transesterication of oil with supercritical methanol Although the FAME yield may increase with an increase of ex-
cess methanol, the cost of recovery of unreacted methanol should
The overall transesterication of triglyceride (rened sunower become higher simultaneously. To nd a favorable molar ratio of
oil in this study) with methanol is simply expressed as methanol to oil is also an important factor to reduce the BD pro-
k1 duction cost. From the experimental results, we found that the
Triglyceride 3Methanol () 3FAME Glycerol 2
A B k2 C D FAME yield reaches to about a constant (0.43) as hB0 are greater
where A to D are triglyceride, methanol, fatty acid methyl esters, than 25 and at 553.2 K, 10.0 MPa, hE0 = 0.1, and s  30 min. It ap-
and glycerol, respectively. The symbols k1 and k2 are the rate con- pears that conducting the transesterication with hB0 = 25 could
stants of forward and backward reactions, respectively. Large excess be a preferable choice.
amount of methanol is needed for favorable the reaction toward the The molar ratio of carbon dioxide to methanol (hE0) is another
FAME formation. Moreover, CO2 (denoted as E) was used as a co- process variable. Imahara et al. (2009) claimed that a negative ef-
solvent to improve the mutual solubility between triglycerides fect on biodiesel production was found as increasing the molar ra-
and methanol. The experiments were conducted at various volu- tio of CO2 to methanol, whereas Han et al. (2005) and Trentin et al.
metric ow rates of the feeds, temperatures from 553.2 to (2011) reported an opposite trend in the presence of CO2. Imahara
593.2 K, pressures from 10.0 to 25.0 MPa, the molar ratios of meth- et al. (2009) indicated that 0.50 of FAME yield was achieved with
anol to oil (hB0) from 20 to 60, and the molar ratio of carbon dioxide hB0 = 42 and hE0 = 0.1 at 543.2 K and 10.0 MPa in 15 min by using
to methanol (hE0) to be 0.1. The residence time (s) in the tabular a batch reactor. In this study, 0.32 of FAME yield was obtained at
reactor was calculated from similar experimental conditions by using the tubular reactor. From
other references, 0.90 of yield was obtained with hB0 = 24 at
s V=v 0 3
573.2 K and 19.0 MPa in 10 min without adding carbon dioxide
with (Bunyakiat et al., 2006). Hawash et al. (2009) observed 1.00 of
v 0 W inlet =qmix;at reaction condition 4 FAME yield in 4 min with hB0 = 43 at 593.2 K and 8.4 MPa without
3
adding carbon dioxide. These results reveal that the presence of
where V is the free-volume in the reactor (about 50 cm , exclude CO2 is unable to improve the FAME yield of the transesterication.
the volume of stainless steel tailings packing), m0 is the volumetric It is may be attributed that the dilution of the concentration of the
ow rate of the feed stream under the experimental conditions, reactants due to the addition of CO2.
Winlet is the total mass ow rate of the feed mixture containing
oil, methanol, and carbon dioxide, and qmix is the density of the feed 3.1.3. Temperature effect
mixture at reaction temperature and pressure (NIST Chemistry Imahara et al. (2008) revealed that biodiesel may decompose
WebBook, 2012; Velez et al., 2010). Table 2 reports the experimen- while reaction temperatures are higher than 623.2 K. To avoid
tal results of transesterication of rened sunower oil with super- the decomposition of product, all the experiments in the present
critical methanol in the presence of CO2. study were operated at temperatures below 623.2 K, from 553.2
This new apparatus was validated by comparing our test run to 593.2 K. The temperature effect on the FAME yield can be found
data with literature value. The FAME yield was found to about from the experimental results of series runs of 35. In these series
0.60 at 573.2 K, 20.0 MPa, hB0 = 42, s = 40 min in the absence of car- runs, the following operating conditions were xed: P = 10.0 MPa,
bon dioxide, while the yield of about 0.55 was reported by Minami hB0 = 25, hE0 = 0.1. Fig. 2 presents the FMAE yield varying with tem-
and Saka (2006) at similar operating conditions. The agreement is perature over the entire range of residence time. As the reaction
reasonably well. This apparatus was further used in the subsequent takes place at 539.2 K, the FAME yield reaches a plateau, around
experiments for investigating the effects of pressure, feed compo- 0.700.72, as s longer than 23 min. The graph also illustrates that
sition, and temperature on the FAME yield of the transesterica- the transesterication rate increases with increasing reaction tem-
tion of the oil with supercritical methanol. perature at a given s.

3.1.1. Pressure effect 3.2. Esterication of oleic acid with methanol

Warabi et al. (2004) and Bunyakiat et al. (2006) reported that
the pressure effect on FAME yield is insignicant using a tubular The esterication of oleic acid with methanol is given by Eq. (5),
reactor, while Silva et al. (2010) found that pressure had a positive
k3
effect on ethyl esters yield by using a microtube reactor. To inves- Oleic acid Methanol () Methyl oleate Water 5
F G k4 I
tigate the pressure effect on the FAME yield, the experiments were H

Table 2
Experimental conditions and range of FAME yield of the transesterication of sunower oil with methanol using CO2 as a co-solvent.

Series of runs T/K P/MPa hB0 hE0 Range of s/min Range of FAME yield
1 553.2 10.025.0 42 0.1 18.6 0.280.32
2 553.2 10.0 2060 0.1 29.530.4 0.320.43
3 553.2 10.0 25 0.1 22.768.1 0.360.56
4 573.2 10.0 25 0.1 22.768.1 0.490.70
5 593.2 10.0 25 0.1 9.868.1 0.470.72

Please cite this article in press as: Tsai, Y.-T., et al. Biodiesel production with continuous supercritical process: Non-catalytic transesterication and ester-
ication with or without carbon dioxide. Bioresour. Technol. (2013), http://dx.doi.org/10.1016/j.biortech.2012.12.157
 Y.-T. Tsai et al. / Bioresource Technology xxx (2013) xxxxxx 5

1 Table 4
Comparison of FAME yield from sunower oil with that from waste cooking oil at
specic experimental conditions.

Sunower oil WCO

0.8 T/K 573.2 573.2
P/MPa 20.0 10.0
hB0 (methanol to oil) 42 25
s/min 40 4
FAME yield 0.60 0.65
0.6
FAME Yield

1
0.4

553.2 K 0.8
0.2 573.2 K
593.2 K
Calc.
0.6
0

XF
0 10 20 30 40 50 60 70 80
(min)
0.4
Fig. 2. Temperature effect on FAME yield of the transesterication conducting at
hB0 = 25, hE0 = 0.1, and P = 10.0 MPa.

533.2 K
0.2
where F to I represent oleic acid, methanol, methyl oleate, and 513.2 K
water, respectively. The variables k3 and k4 are the rate constants 493.2 K
of the forward and the backward reactions, respectively. The exper- Calc.
iments of the esterication were conducted at 10.0 MPa and tem-
0
peratures from 493.2 to 533.2 K with methanol to oleic acid ratio 0 10 20 30 40
from 2 to 5. Table 3 summarizes the experimental results of the
esterication of oleic acid with methanol, runs 610, where CF0 is (min)
the concentration of oleic acid in the feed stream, XF is the conver-
Fig. 3. Temperature effect on the conversion of oleic acid of the esterication
sion of oleic acid, and XFe is the equilibrium conversion of oleic acid. conducting at hG0 = 5 and 10.0 MPa.
(see Table 4)

3.2.1. Temperature effect dence time. The equilibrium conversion of oleic acid increases from
The effect of temperature on the conversion of oleic acid (XF) is 0.93 to 0.97 with hB0 increasing from 2 to 5. It is feasible for the
illustrated in Fig. 3. It shows that the reaction rate, conversion of methyl oleate production as methanol is in large excess.
oleic acid, and equilibrium conversion all increase with increasing
temperature when other operating conditions are xed. In compar-
3.2.3. The effect of phase of methanol
ison with the transesterication, the reaction rate of esterication
The effect of phase of methanol on the kinetic behavior of the
is much faster and the equilibrium conversion can be as high as
esterication was also investigated. Since the critical point of
0.97. Also, the esterication of oleic acid with methanol is an endo-
methanol is at 512.6 K and 8.09 MPa, methanol is at the condition
thermic reaction.
higher than its critical condition as the reaction operated at 533.2 K
and 10.0 MPa, whereas the other two runs were conducted at tem-
3.2.2. Feed composition effect peratures lower than methanols critical temperature. The equilib-
The feed composition effect is shown in Fig. 4. The conversion of rium conversions are about 0.97 at different reaction temperatures,
oleic acid generally increases with increasing hB0 at the same resi- while Fig. 3 shows that the reaction rate and the time to reach

Table 3
Experimental conditions and results of the esterication of oleic acid with methanol and the reaction of WCO with methanol.

Series of runa T/K P/MPa hG0 103 CF0/mol cm3 Range of s/min Range of XF XFe
6 493.2 10.0 5 1.473 3.733.4 0.550.93 0.97
7 513.2 10.0 5 1.346 3.229.2 0.660.95 0.97
8 533.2 10.0 5 7.479 1.715.3 0.720.97 0.97
9 513.2 10.0 2 1.892 3.531.8 0.680.90 0.93
10 513.2 10.0 3 1.668 3.230.8 0.660.93 0.96
11 573.2 10.0 25b 0.108 1.919.0 0.350.90c 0.90d
a
Runs 610: esterication of oleic acid with methanol; run 11: reaction of waste cooking oil with methanol.
b
Methanol to WCO ratio.
c
FAME yield.
d
Maximum FAME yield.

Please cite this article in press as: Tsai, Y.-T., et al. Biodiesel production with continuous supercritical process: Non-catalytic transesterication and ester-
ication with or without carbon dioxide. Bioresour. Technol. (2013), http://dx.doi.org/10.1016/j.biortech.2012.12.157
6 Y.-T. Tsai et al. / Bioresource Technology xxx (2013) xxxxxx

1 dF C =v 0
rC 7
ds
where s has been dened in Eq. (3). Integration of Eq. (7) yields
Z FC
0.8 dF C
s 8
0 v 0 rC
with
0.6 rC k1 C A C 3B  k2 C 3C C D 9
where Ci is the concentration of component i. The yield of FAME is
XF

calculated by the following equation:

0.4 W FAME;out C C;out  MW C
Yield 10
B0 = 5 W TG0 C A0  MW A
where WFAME,out is the mass ow rate of FAME at the outlet, WTG0 the
B0 = 3
0.2 mass ow rate of triglyceride in the feed, Cc,out the concentration of
B0 = 2 FAME at the outlet, CA0 the concentration of triglyceride in the feed
Calc. and MWi the molecular weight of component i. The experimental
data were correlated with Eqs. (8)(10) with the aid of MATLAB.
0 The optimal values of the kinetic parameters were determined by
0 10 20 30 40 minimization of the following objective function:
(min)
1X jYieldexp t
N
k  Yieldcalc
k j
Yield AARD 11
Fig. 4. Feed composition (hG0) effect on the conversion of oleic acid of the N k1 Yieldexp
k
t

esterication conducting at T = 513.2 K and P = 10.0 MPa.

where AARD is average absolute relative deviation and is the num-

equilibrium conversion are substantially faster at the supercritical ber of data points. The correlated results are shown in Table 5, indi-
condition of methanol. At subcritical condition of methanol, the cating that the rate constant of backward reaction is much greater
residence time should be longer than 1 h to approach the equilib- than that of forward constant, thus the equilibrium constant (Kx)
rium conversion. However, the esterication reaches to equilib- is less than unity. Since the equilibrium constant decreases with
rium with the residence time about 15 min. increasing temperature, the transesterication is an exothermic
reaction.
3.3. Transesterication and esterication of WCO with methanol The calculated results are compared with the experimental val-
ues in Fig. 2. Generally, the agreement is satisfactory over the en-
The transesterication of oils with methanol and the esterica- tire experimental range. The rate constants of the forward and
tion of fatty acids with methanol will be taking place simulta- backward reactions conducting at hB0 = 25, hE0 = 0.1, and
neously if the oil feedstock contains a certain amount of FFA. To P = 10.0 MPa are expressed as the Arrhenius form:
investigate the inuence of feedstock in presence of FFA, rened
sunower oil and WCO were used as a reactant, respectively. The 10284:9
k1 1:936 104 exp 12
experimental results show that the FAME yield of 0.65 can be T
achieved at T = 573.2 K, P = 10.0 MPa, hG0 = 25, and s = 4 min using
the WCO as feedstock. It took much longer time (s = 40 min) to 20198:7
k2 1:785 1014 exp 13
reach the same yield when rened sunower oil was used as feed- T
stock. The FAME yield of 0.90 can be achieved at s = 13 min by where the units of ki and T are m9 kmol3 min1 and K, respectively.
using WCO as feedstock, indicating that the presence of FFA can According to the above correlation, the activation energies of the
enhance the production rate of FAME through the non-catalytic transesterication (E0,f) is about 85.5 kJ mol1 and the effective mo-
reactions with supercritical methanol. With this non-catalytic lar heat of reaction is about 82.4 kJ mol1 (i.e., an exothermic reac-
supercritical method, we do not need to remove FFA and water tion). Kusdiana and Saka (2001) reported that the activation energy
from the feedstock before entering the reactor. It allowed us to of 69.2 kJ mol1 was obtained at temperatures from 473.2 to
convert low grade feedstock to FAME efciently. 573.2 K, and Wang and Yang (2007) also reported the activation en-
ergy of 92.0 kJ mol1 for transesterication of soybean oil with
3.4. Correlation of kinetics data methanol at 473.2533.2 K. Comparing with those results, we found
that the activation energy of the forward reaction determined from
3.4.1. Transesterication this study agrees with the literature values reasonably well.
A power law model was used for correlating the non-catalytic
kinetic data of the transesterication of the oil with methanol. 3.4.2. Esterication
The mass-balance equation for FAME (C) around the continuous The non-catalytic kinetic data of esterication of oleic acid with
tubular reactor is given by the following equation: methanol were also correlated with a power law model. The mass-
balance equation for oleic acid (F) around the tubular reactor is gi-
dF C
rC 6 ven by Eq. (14).
dV
Z XF
where V is the free-volume in the reactor, FC the molar ow rate of dX F
s 14
FAME, and rC the molar production rate of FAME per unit volume. 0 rF
We assumed that the density change of the reacting mixture is neg-
ligible through the whole reactor, Eq. (6) becomes with  r F k3 C F C G  k4 C H C I 15

Please cite this article in press as: Tsai, Y.-T., et al. Biodiesel production with continuous supercritical process: Non-catalytic transesterication and ester-
ication with or without carbon dioxide. Bioresour. Technol. (2013), http://dx.doi.org/10.1016/j.biortech.2012.12.157
 Y.-T. Tsai et al. / Bioresource Technology xxx (2013) xxxxxx 7

Table 5
Correlated results of the kinetic data of transesterication of sunower oil with methanol.

T/K hB0 hE0 104 k1/m9 kmol3 min1 k2a/m9 kmol3 min1 102 Kx 102 Yield AARDb
553.2 25 0.1 1.797 0.0257 0.70 4.7
573.2 25 0.1 2.558 0.0825 0.31 4.5
593.2 25 0.1 6.339 0.3019 0.21 3.4
a
k2 k1 =K x .
P jYieldexpt Yieldcalc
k j
b
Yield AARD N1 N
k1
k
Yieldexpt
, where N is the number of data points.
k

where k3 and k4 are the rate constants of forward and backward esterication is an endothermic reaction and the reaction rate in
reactions, respectively. The subscripts G, H, I denote methanol, the supercritical state of methanol is much faster than in the sub-
methyl oleate, and water, respectively. Expressing the rate equation critical state. Alenezi et al. (2010) reported the activation energy is
in terms of conversion of oleic acid (XF) and integrating Eq. (14) about 72 kJ mol1, which were determined their esterication data
yields at temperatures from 523 to 593 K and 10 MPa with hG0 = 7 by
  using a batch reactor. This value is obviously lower than our result
X Fe q X F  X Fe
k3 1=K x  1qC F0 s ln 16 in supercritical state of methanol, but greater than that in subcrit-
X Fe  X F q  X Fe
ical state of methanol. The discrepancy may be attributive to the
where different feedstock sources, the molar ratios of methanol to FFA,
q and hydrodynamics due to using different types of reactor.
K 2x hG0  12 4K x hG0
q 17
1  Kx 4. Conclusions
The rate constant k3 and the equilibrium constant Kx were
determined from the kinetic data correlation with Eq. (16) by using The reaction rate and FAME yield of the transesterication in-
MATLAB as an optimization tool and the AARD of XF as the objec- creased with increasing temperature but insensitive to pressure.
tive function. The correlated results are reported in Table 6. The The presence of CO2 cannot enhance the reaction rate and increase
calculated results are compared with experimental values in Figs. 3 the yield of FAME. The esterication rate increases with increasing
and 4. The power law model correlated the kinetic data to AARDs temperature and the equilibrium conversion of oleic acid, increas-
no greater than 3%. As can be seen from Table 6, the rate constant ing from 0.93 to 0.97 at 513.2 K and 10.0 MPa. The kinetic data
of the forward reaction k3 jumps about vefold from subcritical to were correlated satisfactorily with a power law model. By applying
supercritical temperature of methanol. The rate constants of the the supercritical technique for producing FAME from WCO, higher
forward and the backward reactions operating at hG0 = 5 and FAME yields and faster reaction rate have been achieved, even
10.0 MPa can be expressed in the Arrhenius form. The tempera- without pretreatment of the feedstock before entering the reactor.
ture-dependent expressions in the supercritical and the subcritical
regions are given below:
Acknowledgement
 
6581:8
k3;sub 1:740 107 exp 18 The authors gratefully acknowledge the nancial support from
T
National Science Council, Taiwan, through Grant No. NSC 97-
  2221-E011-049-MY3.
21282:6
k3;sup 4:812 1019 exp 19
T
  References
3870:9
k4;sub 2:266 104 exp 20
T Alenezi, R., Leeke, G.A., Winterbottom, J.M., Santos, R.C.D., Khan, A.R., 2010.
Esterication kinetics of free fatty acids with supercritical methanol for
  biodiesel production. Energy Convers. Manage. 51, 10551059.
15914:6 Bertoldi, C., da Silva, C., Bernardon, J.P., Corazza, M.L., Filho, L.C., Oliveira, J.V.,
k4;sup 3:527 1014 exp 21 Corazza, F.C., 2009. Continuous production of biodiesel from soybean oil in
T
supercritical ethanol and carbon dioxide as cosolvent. Energy Fuels 23, 5165
The activation energy of the esterication in subcritical state of 5172.
Bunyakiat, K., Makmee, S., Sawangkeaw, R., Ngamprasertsith, S., 2006. Continuous
methanol (E0,f) is about 54.7 kJ mol1 and the effective molar heat production of biodiesel via transesterication from vegetables oils in
of reaction is about 22.5 kJ mol1 (i.e., an endothermic reaction) supercritical methanol. Energy Fuels 20, 812817.
while those in supercritical state are about 176.9 and 44.6 kJ mol1 Cheng, J.J., Timilsina, G.R., 2011. Status and barriers of advanced biofuel
technologies: a review. Renew. Energy 36, 35413549.
(also an endothermic reaction), respectively. It indicates that the de Boer, K., Bahri, P.A., 2011. Supercritical methanol for fatty acid methyl ester
production: a review. Biomass Bioenergy 35, 983991.
Demirbas, A., 2006. Biodiesel production via non-catalytic SCF method and
Table 6 biodiesel fuel characteristics. Energy Convers. Manage. 47, 22712282.
Correlated results of the kinetic data of esterication of oleic acid with methanol. Dippolito, S.A., Yori, J.C., Iturria, M.E., Pieck, C.L., Vera, C.R., 2007. Analysis of a two-
step, noncatalytic, supercritical biodiesel production process with heat
T/K hG0 k3/ k4a/ Kx 102 XF recovery. Energy Fuels 21, 339346.
cm3 mol1 min1 cm3 mol1 min1 AARDb Han, H., Cao, W., Zhang, J., 2005. Preparation of biodiesel from soybean oil using
supercritical methanol and CO2 as co-solvent. Proc. Biochem. 40, 31483151.
493.2 5 27.8 8.8 3.2 3.0
Hawash, S., Kamal, N., Zaher, F., Kenawi, O., El Diwani, G., 2009. Biodiesel fuel from
513.2 5 46.8 12.0 3.9 2.0
Jatropha oil via non-catalytic supercritical methanol transesterication. Fuel 88,
533.2 5 221.8 38.3 5.8 0.8 579582.
513.2 2 104.8 15.2 6.9 2.0 Imahara, H., Minami, E., Hari, S., Saka, S., 2008. Thermal stability of biodiesel in
513.2 3 72.1 13.8 5.2 1.9 supercritical methanol. Fuel 87, 16.
a
Imahara, H., Xin, J., Saka, S., 2009. Effect of CO2/N2 addition to supercritical
k4 k3 =K x . methanol on reactivities and fuel qualities in biodiesel production. Fuel 88,
P jX F expt X F calc
k j
b
X F AARD N1 N k1
k
X expt
, where N is the number of data points. 13291332.
F k

Please cite this article in press as: Tsai, Y.-T., et al. Biodiesel production with continuous supercritical process: Non-catalytic transesterication and ester-
ication with or without carbon dioxide. Bioresour. Technol. (2013), http://dx.doi.org/10.1016/j.biortech.2012.12.157
8 Y.-T. Tsai et al. / Bioresource Technology xxx (2013) xxxxxx

Juan, J.C., Kartika, D.A., Wu, T.Y., Hin, T.Y.Y., 2011. Biodiesel production from Semwal, S., Arora, A.K., Badoni, R.P., Tuli, D.K., 2011. Biodiesel production using
Jatropha oil by catalytic and non-catalytic approaches: an overview. Bioresour. heterogeneous catalysts. Bioresour. Technol. 102, 21512161.
Technol. 102, 452460. Sharma, Y., Singh, B., Korstad, J., 2011. Latest developments on application of
Keera, S.T., El Sabagh, S.M., Taman, A.R., 2011. Transesterication of vegetable oil to heterogeneous basic catalysts for an efcient and eco friendly synthesis of
biodiesel fuel using alkaline catalyst. Fuel 90, 4247. biodiesel: a review. Fuel 90, 13091324.
Kusdiana, D., Saka, S., 2001. Kinetics of transesterication in rapeseed oil to Shieh, C.J., Liao, H.F., Lee, C.C., 2003. Optimization of lipase-catalyzed biodiesel by
biodiesel fuel as treated in supercritical methanol. Fuel 80, 693698. response surface methodology. Bioresour. Technol. 88, 103106.
Kusdiana, D., Saka, S., 2004. Effects of water on biodiesel fuel production by Silva, C., Castilhos, F., Oliveira, J.V., Filho, L.C., 2010. Continuous production of
supercritical methanol treatment. Bioresour. Technol. 91, 289295. soybean biodiesel with compressed ethanol in a microtube reactor. Fuel
Lotero, L., Liu, Y., Lopez, D.E., Suwannakarn, K., Bruce, D.A., Goodwin Jr., J.G., 2005. Process. Technol. 91, 12741281.
Synthesis of biodiesel via acid catalysis. Ind. Eng. Chem. Res. 44, 53535363. Song, E.S., Lim, J.W., Lee, H.S., Lee, Y.W., 2008. Transesterication of RBD palm oil
Ma, F., Hanna, M., 1999. Biodiesel production: a review. Bioresour. Technol. 70, 1 using supercritical methanol. J. Supercrit. Fluids 44, 356363.
15. Tan, K.T., Lee, K.T., 2011. A review on supercritical uids (SCF) technology in
Macario, A., Giordano, G., Onida, B., Cocina, D., Tagarelli, A., Giuffre, A.M., 2010. sustainable biodiesel production: potential and challenges. Renew. Sustainable
Biodiesel production process by homogeneous/heterogeneous catalytic system Energy Rev. 15, 24522456.
using an acidbase catalyst. Appl. Catal. A: Gen. 378, 160168. Trentin, C.M., Lima, P.A., Alkimin, I.P., Silva, C., Castilhos, F., 2011. Continuous
Marulanda, V.F., Anitescu, G., Tavlarides, L.L., 2010. Biodiesel fuels through a production of soybean biodiesel with compressed ethanol in a microtube
continuous ow process of chicken fat supercritical transesterication. Energy reactor using carbon dioxide as co-solvent. Fuel Process. Technol. 92, 952958.
Fuels 24, 253260. Velez, A., Hegal, P., Mabe, G., Brignole, E.A., 2010. Density and conversion in
Minami, E., Saka, S., 2006. Kinetics of hydrolysis and methyl esterication for biodiesel production with supercritical methanol. Ind. Eng. Chem. Res. 49,
biodiesel production in two-step supercritical methanol process. Fuel 85, 2479 76667670.
2483. Wang, L., Yang, J., 2007. Transesterication of soybean oil with nano-MgO or not in
Mittelbach, M., Remschmidt, C., 2006. Biodiesel: The Comprehensive Handbook. supercritical and subcritical methanol. Fuel 86, 328333.
Martin Mittelbach, Vienna. Wang, R., Zhou, W.W., Hanna, M.A., Zhang, Y.P., Bhadury, P.S., Wang, Y., Song, B.A.,
NIST Chemistry WebBook. Isothermal properties for carbon dioxide. NIST Standard Yang, S., 2012. Biodiesel preparation, optimization, and fuel properties from
Reference Database No. 69, March 29, 2012 updated, National Institute of non-edible feedstock, Datura stramonium L.. Fuel 91, 182186.
Standards and Technology, USA. Available from: <http://webbook.nist.gov/ Warabi, Y., Kusdiana, D., Saka, S., 2004. Biodiesel fuel from vegetable oil by various
chemistry/>). supercritical alcohols. Appl. Biochem. Biotechnol. 115, 793801.
Saka, S., Kusdiana, D., 2001. Biodiesel fuel from rapeseed oil as prepared in Yusuf, N.N.A.N., Kamarudin, S.K., Yaakub, Z., 2011. Overview on the current trends
supercritical methanol. Fuel 80, 225231. in biodiesel production. Energy Convers. Manage. 52, 27412751.

Please cite this article in press as: Tsai, Y.-T., et al. Biodiesel production with continuous supercritical process: Non-catalytic transesterication and ester-
ication with or without carbon dioxide. Bioresour. Technol. (2013), http://dx.doi.org/10.1016/j.biortech.2012.12.157