Journal of Membrane Science 496 (2015) 242249

Contents lists available at ScienceDirect

Journal of Membrane Science

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

Inuence of oil quality on biodiesel purication by ultraltration

Maria Carolina Srgi Gomes a,n, Pedro Augusto Arroyo b, Nehemias Curvelo Pereira b
a
Federal University of Technology of Paran-UTFPR, Rua Marclio Dias, 635, CEP 86812-460 Apucarana, PR, Brazil
b
Department of Chemical Engineering, State University of Maring, Av. Colombo 5790, Bloco D90, CEP 87020-900 Maring, PR, Brazil

art ic l e i nf o a b s t r a c t

Article history: In the present study, biodiesel was produced by ethylic transesterication of soybean and canola oils
Received 19 May 2015 using sodium hydroxide as a catalyst. It was evaluated the inuence of oil quality on the biodiesel and
Received in revised form glycerol separation by ultraltration. The experiments were carried out with tubular a-Al2O3/TiO2
31 July 2015
membranes with average pore diameter of 0.05 mm and 20 kDa, varying the transmembrane pressure
Accepted 4 September 2015
and the concentration of the feed mixture. The comparison among the use of degummed soybean oil,
Available online 8 September 2015
rened soybean oil, crude canola oil, and rened canola oil, demonstrated that free fatty acid presented
Keywords: in the oils inuence the formation of droplets containing glycerol. The separation was efcient when
Ultraltration reaction mixture was produced from degummed soybean oil and crude canola oil, both with a higher
Ceramic membranes
acidity value. The highest free fatty acid content in the crude canola oil, not only favored the formation of
Transesterication
a dispersed phase containing glycerol, which was retained by the membrane, but also resulted in the
Biodiesel
Glycerol lowest ux decline rates. The ultraltration was efcient in removing glycerol, since the highest glycerol
content in the permeate was 0.013 wt%. This novel rening process of biodiesel showed the advantage of
not requiring previous decantation to separate the two phases obtained after transesterication and the
reduction in the amount of water used in the washing steps. The properties of the biodiesel produced,
which were evaluated, meet the ANP biodiesel standards required for marketing.
& 2015 Elsevier B.V. All rights reserved.

1. Introduction
Biodiesel is considered a renewable fuel, as vegetable oil and
animal fat are the main raw materials for its production. It has
higher ash point than mineral diesel, ensuring greater safety in
use and also presents appropriate viscosity for burning in diesel
engines. It is biodegradable, non-toxic and has excellent lubricity,
providing longer life to components of engines [1].
The most common method for producing biodiesel is the
transesterication, a reaction between a triacylglycerol, from ve-
getable oils or animal fats, and a short chain alcohol producing
alkyl esters of fatty acids and glycerol as a co-product [2,3]. Bio-
diesel has similar physicochemical properties to that of mineral
diesel and can be used in engines without modication [3].
The potential use of a raw material in biodiesel production
depends on several factors, such as availability, cost, storage
properties and performance as fuel [4]. Given the high biodi-
versity, large territory, climate diversity and soil conditions, Brazil
contains different oil sources for biodiesel production including
soybean, canola, castor, babassu, cotton, palm and sunower [5].
Since Brazil has a well stablished soybean production, presently
n
Corresponding author.
E-mail address: mariagomes@utfpr.edu.br (M.C.S. Gomes).
most of biodiesel produced in the country used soybean oil [5,6].
Canola presents a high percentage of seed oil, 3440%, which is
approximately twice the value found in soybeans and correspond
to 84% of all raw materials used in global biodiesel production
[7,8].
Regarding the alcohol used in the process, in Brazil, the use of
ethyl route has a strategic importance, since the availability of raw
materials and technology allows an economically viable produc-
tion of ethanol by fermentation processes, resulting in a cheaper
product than the methanol. In biodiesel production, the use of
vegetable oils and ethanol derived from sugarcane makes the
process completely independent from petroleum, providing en-
vironmental benets and generating a socio-economic develop-
ment program [9,10].
After transesterication, the nal mixture is mainly composed
of alkyl esters of fatty acids, residual alcohol, glycerol, catalyst,
mono-, di- and triglycerides. These and other contaminants in the
biodiesel can cause operational and environmental problems and
must be separated so that esters can be used as fuel [9,11]. The
quality of B100 biodiesel in Brazil is specied by the National
Agency of Petroleum, Natural Gas and Biofuels (ANP), based on the
American ASTM D6751 and European EN14214 standards, with
some modications to meet the Brazilian raw material require-
ments [12].
One of the most important parameters for quality control of

http://dx.doi.org/10.1016/j.memsci.2015.09.004
0376-7388/& 2015 Elsevier B.V. All rights reserved.
 M.C.S. Gomes et al. / Journal of Membrane Science 496 (2015) 242249
biodiesel is the amount of free glycerin, which has a maximum
allowed limit of 0.02%. A high concentration of free glycerin can
result from the separation of glycerin, causing problems during
storage and in the fuel injection system [9,13].
Thus, besides controlling all the parameters inuencing the
reaction of biodiesel production, the stage of glycerol separation is
very important to achieve a quality product, free of impurities and
no risk of corrosion to the engine. Conventional purication pro-
cedures use a large volume of water, which varies depending on
the impurities present, and provides the generation of a large
amount of efuent that must be properly treated and disposed
[14].
Studies in the literature [1519] indicate that distinct methods
have been applied for glycerol removal and biodiesel purication.
Among the alternative methods of separation, the process with
membranes has many advantages and good prospects for use in
the separation and purication of biodiesel. The use of membranes
can provide high quality and purity biodiesel, and environmental
and economic advantages for reducing the amount of water used
and eliminating the use of adsorbents. Apart from the reduction of
costs related to efuent treatment, studies indicate that the use of
membranes in the processing of biodiesel provides reduced power
consumption, enabling the application of this process in industrial
scale [20].
Regarding the use of membranes for biodiesel production, the
published studies have focused on the use of membrane reactors
to improve conversion into esters [2123] and the use of mem-
branes in the purication step, after separation by decantation
[18,2428]. Direct use of micro- and ultraltration after transes-
terication for the separation of phases, without previous decan-
tation, was studied in previous works developed by our research
group. The rst experiments [29] were carried out with micro-
ltration membranes using synthetic mixtures of biodiesel, gly-
cerol, and ethanol. They were evaluated the effects of the mem-
brane porosity, transmembrane pressure, and the ethanol con-
centration in the mixture. The results indicated the inuence of
ethanol on behavior of the emulsion and the potential of applying
ceramic membranes in the separation of glycerol and biodiesel.
Afterwards [30], the experiments were performed using the
reactional mixture produced by ethylic transesterication of de-
gummed soybean oil. Considering the emulsion behavior of the
mixture produced in the transesterication of degummed soy-
bean oil, it was developed a methodology of addition of acidied
water, aiming to destabilize this emulsion and improve the re-
tention of glycerol by the membrane. The results showed the key
role of both ethanol and water in the formation of agglomerates in
the dispersed phase, since the glycerol retention after adding
water was signicant, with glycerol mass content in the permeate
below 0.02%, indicating the efciency of the used methodology.
Continuing the study [31], it was evaluated the inuence of the
amount of acidied water added on the separation of glycerol and
biodiesel using micro and ultraltration ceramic membranes. The
results showed that the amount of acidied water added inu-
ences the emulsion properties and, consequently, the distribution
of agglomerate size containing the glycerol.
The current work aimed at improving knowledge on the be-
havior of ultraltration with ceramic membranes in separating
glycerol from biodiesel using different feedstock. To this end, it
was evaluated the inuence of oil quality on ultraltration of the
reaction mixture produced after ethyl esterication of vegetable
oils with different characteristics. The separation of glycerol was
performed using ultraltration membranes, and the best operating
condition was evaluated in terms of permeate ux and quality of
the product.
243
2. Experimental section
2.1. Production of biodiesel
The reaction mixtures needed for the experiments were pre-
pared by alkaline ethyl transesterication of vegetable oils. They
were used degummed soybean oil and crude canola oil donated by
Cocamar (Maring, Paran State, Brazil), and commercial rened
soybean and canola oils. The anhydrous ethanol (99.4% purity) was
donated by Cocaf, Astorga, Paran State, Brazil and sodium hy-
droxide (NaOH) was purchased from Biotec.
The previous results of the study of the yield of esters as a
function of reaction parameters [30] indicate that the reaction
temperature for obtaining a high ester yield depends on the
acidity of the oil. In this way, when used rened oils, the reaction
temperature was 45 C, and when used crude and degummed oils,
the transesterication was carried out at 30 C. For all oils, the oil:
alcohol molar ratio used was 1:7.5 and the amount of catalyst was
1% of mass of oil. The reaction was prepared in a 2 L batch reactor
and the reaction time was 1 h. The reactor contents were mixed
using a mechanical stirrer.
2.2. Ultraltration runs
The experimental equipment consisted of a micro- and ultra-
ltration pilot unit UF NETZSCH, Pomerode, Santa Catarina State,
Brazil, model 027.06-1C1/07-0005/AI, operating in cross-ow
conditions. A detailed description of the experimental unit was
presented in a previous study [30].
The hydrophilic ceramic membranes used in the experiments
were made of tubular -Al2O3/TiO2 (Shumacher GmbHTi 01070),
250 mm long, 7 mm in diameter, and 0.005 m2 ltration area,
purchased from Andritz, Pomerode, Santa Catarina State, Brazil.
The tests were carried out with ultraltration membranes of
0.05 mm and 20 kDa. All ultraltration experiments were per-
formed at 50 C.
In the rst step, ultraltration of the mixture obtained from
rened canola oil was evaluated. For all tests, before ultraltration
of the mixture, the addition of acidied water (0.5% HCl) was
evaluated in the mass concentrations varying from 0% to 20% in
relation to the total mixture mass, under transmembrane pressure
of 0.52.0 bar. As previously discussed [31], addition of water af-
fords the formation of dispersed phase containing water, glycerol,
catalyst, salts and other water-soluble substances, distinct from
the continuous phase rich in ethyl esters and unreacted oil.
Next, ultraltration experiments were run with reaction mix-
tures produced with rened soybean, degummed soybean, rened
canola and raw canola oils. The experiments were performed only
with the 0.05 mm membrane, under pressure of 1.0 bar, with prior
addition of 10% acidied water.
Finally, it was evaluated the separation of glycerol from the
mixture produced with raw canola oil. In these experiments were
used the membranes of 0.05 mm and 20 kDa, under pressures of
1.0, 2.0 and 3.0 bar and water concentration of 10%.
Approximately four liters of the mixture were poured into the
feed tank of the module for each run. After heating to 50 C, the
mixture was pumped into the membrane and the pressure was set
to begin the permeate ux. All the runs were undertaken with a
ow rate of 700 L/h, corresponding to a tangential velocity of
about 8 m/s. The permeate was collected and the concentrate was
completely recirculated to the feed tank.
The performance of the glycerol separation process by ceramic
membranes were characterized in terms of glycerol rejection,
stabilized permeate ux, percentage of ux decline and ux with
pure water after membrane cleaning.
The reaction mixture obtained after adding acidied water was
244 M.C.S. Gomes et al. / Journal of Membrane Science 496 (2015) 242249
Table 1
Physicochemical characteristics of the vegetable oils used.
Vegetable oil Viscosity at 40 C (mm2/s)
Degummed soybean 32.55
Rened soybean 32.01
Crude canola 32.22
Rened canola 36.39
directly fed to the micro- and ultraltration module, substituting
the steps of settling and washing with water by micro- or ultra-
ltration. Thus the use of membranes for glycerol separation en-
ables the reduction in the number of steps in relation to the
conventional treatment, in addition to a lower consumption of
water.
2.3. Analytical methods
The viscosity at 40 C was determined with a Brookeld model
DV-III digital rheometer. A 25-mL pycnometer with a coupled
thermometer was used to determine the density at 20 C. The
acidity was determined according to the American Oil Chemical
Society (AOCS). The moisture content was determined by the Karl
Fished method using an Analyzer apparatus, model Moisture
Control KF-1000. The caloric value was determined using a Parr
6200 calorimeter. All analyses were performed in triplicate.
The fatty acid prole of the oil and the quantication of the
ethyl esters were done by gas chromatography with Varian model
CP-3800 coupled with a ame ionization detector (FID) and a
30 mx0.25 mm capillary column specic for fatty acid separation,
BP-X70. The carrier gas used was helium, in a split rate of 1:10. The
analysis was performed with a column temperature program
starting at 140 C, heating up to 250 C at 5 C/min. The detector
temperature was kept at 220 C and that of the injector at 260 C.
The internal standard used was 99% methyl tricosanoate, pur-
chased from Sigma-Aldrich.
The free glycerol content of the permeate was determined by
volumetric method based on the ofcial AOCS methodology for
the analysis of free glycerol in oils and fats (Ca 14-56) [32]. Each
sample was analyzed three times and the averages are reported.
The titration method with sodium periodate allows fast and ac-
curate determination of glycerol in oils and fats at a low cost
compared with chromatographic methods [9,32,33]. When gly-
cerol concentration in the permeate was too high, i.e., when there
was permeation of the dispersed phase through the membrane,
titration method could not be applied and it was only performed a
volumetric quantication of the percentage of the dispersed phase
containing glycerol.
2.4. Membrane cleaning process
Although membrane processes are extensively employed in the
industry, the major disadvantage of their application is the fouling
of the membranes, which imposes the need for frequent cleaning.
Therefore, an improvement of the cleaning procedure may have a
signicant inuence on the overall process efciency [34].
Current membrane cleaning technologies include hydraulic,
chemical, and mechanical methods. Ultrasound is an effective
technique for cleaning a variety of surfaces, which has been used
for cleaning fouled membranes or for increasing permeate ux of
water through membranes [35]. The basic principle of operation is
that ultrasound removes particles from the surface by causing
particle movement in or near membrane [36].
In this work, the membrane cleaning was performed in ultra-
sonic bath (Ultra Cleaner 800, UNIQUE) at 40 kHz, with solutions
Density at 20 C (g/cm3) Moisture (%) Acidity (%)
0.9156 7 0.0007 0.1277 0.004 0.87 70.01
0.91227 0.0007 0.090 7 0.003 0.43 70.05
0.9105 7 0.0007 0.1137 0.004 2.40 70.06
0.91347 0.0007 0.098 7 0.004 0.2470.03
of 1% NaOH and 1% citric acid at 70 C and then with deionized
water at the same temperature. It was evaluated the membrane
hydraulic permeability so that a cleaning parameter could be set.
After each regeneration cycle, the permeate ux was measured
with deionized water, ensuring the reproducibility of the tests.
3. Results and discussion
3.1. Characteristics of the oils
Table 1 lists the physicochemical characteristics of the four
types of oil used in the experiments: degummed soybean oil, re-
ned soybean oil, rened canola oil and crude canola oil.
Vegetable oils exhibited similar density and viscosity and in
accordance with the values in the literature [7,37]. The degummed
soybean oil and crude canola oil showed acidity values of 0.87%
and 2.4%, respectively. These percentages are higher than the re-
commended by Freedman et al. [38] for alkaline transesterica-
tion. Nevertheless, as studies have shown good results for this
reaction with the use of vegetable oils with up to 3% acidity [39]
and given the possibility of reducing the costs of treating the raw
material, oils were not pre-treated. For all oils, the moisture con-
tent was below 0.5%, which is the maximum value recommended
for the alkaline transesterication is not impaired [38].
3.2. Ultraltration in the separation of biodiesel and glycerol pro-
duced by ethyl transesterication of rened canola oil
Ultraltration of the mixture produced from rened canola oil,
whose acidity was the lowest of all evaluated oils, was not efcient
in the separation of glycerol. Considering the stoichiometry of the
reaction, the amount of glycerol in feed stream varied from 6% to
7%, depending on the quantity of water added.
For all the conditions evaluated, the dispersed phase containing
glycerol permeated through the membrane, and the percentage of
this phase in the permeate varied from 1% to 44% (Table 2).
These results are justied by the absence of acidity in the oil,
which prevented the agglomeration of the dispersed phase in the
nal mixture. Saleh et al. [26] studied biodiesel purication using
membrane process and demonstrated that at lower soap con-
centrations, the increasing amount of water promotes the gen-
eration of larger glycerol rich particles. In this present work, soaps
formed in the reaction, when there are free fatty acids in the oil,
along with salts from the neutralization with acidied water,
contribute to the formation of a dispersed phase containing gly-
cerol, which can be retained by the membrane.
Fig. 1 illustrates the samples of permeates in all conditions
evaluated, indicating the presence of two phases. The lower phase
is the dispersed phase containing glycerol, and the upper phase is
formed mainly by ethyl esters.
Permeate ux curves over ltration time, produced from ethyl
transesterication of rened canola oil, using membranes of
0.05 mm and 20 kDa are presented in Figs. 2 and 3, respectively.
It is interesting to observe a ux increasing in the beginning of
the ltration for most of the runs. Probably, until the phase
 M.C.S. Gomes et al. / Journal of Membrane Science 496 (2015) 242249 245

Table 2 450
Concentration of the dispersed phase in the permeate ultraltration of rened
canola oil. 400
1.0 bar - 1%
Sample Membrane Pressure Concentration of Dispersed 350 1.0 bar - 3%
(mm/kDa) (bar) acidied water (% phase (% mass) 1.0 bar - 10%

Permeate Flux (kg/h.m )

mass)

2
300 2.0 bar - 0%
1 0.05 0.5 10 1.7
2.0 bar - 5%
2 0.05 1.0 1 31
250 2.0 bar - 10%
3 0.05 1.0 3 8.4
4 0.05 1.0 10 12 200
5 0.05 2.0 0 34
6 0.05 2 5 18 150
7 0.05 2.0 10 22
8 20 0.5 3 4 100
9 20 0.5 10 5
10 20 1.0 1 3 50
11 20 1.0 10 1
12 20 2.0 0 3 0
13 20 2.0 10 10
0 10 20 30 40 50 60 70 80 90 100
14 20 2.0 20 44
Time (min)
Fig. 2. Permeate ux in the ultraltration of biodiesel from rened canola oil for all
conditions evaluated. 0.05 mm membrane, T 50 C.

0.5 bar - 3%
240 0.5 bar - 10%
1.0 bar - 1%
210 1.0 bar - 10%
2.0 bar - 0%
180 2.0 bar - 10%
Permeate Flux (kg/h.m )
2

2.0 bar - 20%

150

120

90

60

30

0
0 10 20 30 40 50 60 70 80 90 100
Time (min)
8 9 10 11 12
13 14 Fig. 3. Permeate ux in the ultraltration of biodiesel from rened canola oil for all
conditions evaluated. 20 kDa membrane, T 50 C.

According to Fig. 2, the 0.05 mm membrane showed very high

Fig. 1. Samples of ultraltration permeates of the rened canola oil.
initially retained on the membrane surface is removed by the
convective ux and a concentration prole is established, which is
the transient period, there is an increase in the ux. After this
initial period, due to the size of glycerol agglomerates and the
pressure used, permeation occurs through the membrane pores,
and consequently the fouling, which reduces the ux.
ux rates, so that, in some tests, the minimum volume required in
the feed tank was achieved before the default time of 95 min. The
use of 2.0 bar pressure promoted a sharp decline in ux for all
concentrations of water, as well as a high glycerol content in the
permeate (Table 1). Although the lower pressure, 1.0 bar, has
caused a reduction of membrane fouling, there was glycerol per-
meation for the three concentrations of water evaluated, and the
lowest value of glycerol phase in the permeate was 8.4%, with
acidied water concentration of 3%.
Given the same concentration of water, 10%, increased pressure
of 1.0 to 2.0 bar provided increased glycerol content in the
permeate from 12 to 22% in the 0.05 mm membrane and 1 to 10% in
the 20 kDa membrane.
On the other hand, at constant pressure, it is observed that in
0.05 mm membrane, there was an increased retention of glycerol
with higher concentration of water compared to the lowest con-
centration studied. As for the 20 kDa membrane, increased re-
tention of glycerol with increasing concentration of water was
only veried at 1.0 bar pressure. At 2.0 bar pressure, increasing the
concentration of water promoted a reduction of glycerol retention,
246 M.C.S. Gomes et al. / Journal of Membrane Science 496 (2015) 242249
and with the maximum concentration of water evaluated, it was
obtained the highest content of glycerol phase in the permeate,
44%. Compared to other results, it is observed that the addition of
20% water caused a great increase in permeation of glycerol,
probably indicating inversion of the emulsion.
From these results, it was concluded that for the two mem-
branes at the same concentration of water, the increased pressure
favors the permeation of glycerol. The 0.05 mm membrane showed
an average value of glycerol phase in permeate greater than
20 kDa membrane, suggesting that glycerol agglomerates have
sizes near the pore diameter of the 0.05 mm membrane.
As the best retention of glycerol in the permeate was obtained
with the 20 kDa membrane at 1.0 bar pressure and 10% water,
additional tests were performed by reducing the pressure to
0.5 bar. However, even with the reduction of pressure, for the two
water concentrations tested, there was no glycerol retention by the
membrane.
3.3. Inuence of the type of oil used in the separation
To evaluate the inuence of the type of oil used in membrane
separation, ultraltration tests were run with reaction mixtures
obtained from rened and degummed soybean oils and from
crude and rened canola oils. Thus, it was possible to compare the
characteristics of ultraltration of the biodiesel produced from oil
with a higher acidity (degummed soybean and crude canola) and
biodiesel produced from rened oil, which has low acidity. Tests
were performed at 50 C at a pressure of 1.0 bar with prior addi-
tion of 10% acidied water, and the curves of permeate ux over
time are presented in Fig. 4.
According to the results, an increase is observed in the ux for
rened oils in the 10 initial minutes, followed by a constant de-
cline over time. As already discussed previously, glycerol per-
meation provides fouling increasing.
The curve for degummed soybean oil shows a typical behavior
of membrane separation processes, with concentration polariza-
tion and fouling. The fouling is caused by external clogging of
pores, but without glycerol permeation, according to the percen-
tage of glycerol in the permeate listed in Table 3.
In turn, for the crude canola oil, the decline of ux over time is
minimum, indicating that the retained phase does not cause
clogging of pores. The presence of fatty acids in the oil and hence,
soaps and salts formed in the reaction, favor the formation of
Fig. 4. Permeate ux in the ultraltration of biodiesel from soybean and biodiesel
from canola. Membrane of 0.05 mm, P 1.0 bar, 10% acidied water.
Table 3
Comparison of the permeate ux and glycerol concentration of the ultraltration
with different vegetable oils studied.
Raw material Permeate ux (kg/h m2) Glycerol in the permeate (%)
Degummed soybean 69.7 0.0187 0.001
Rened soybean 166.1 12
Crude canola 101.1 0.013 70.003
Rened canola 121.4 12
agglomerates containing glycerol and water-soluble substances.
Such agglomerates are likely much larger than the membrane
pores, thus retained more easily.
Comparing the rened soybean oil with degummed soybean oil
and rened canola oil with crude canola oil, it can be concluded
that the formation of the dispersed phase is favored by the pre-
sence of fatty acids in the oil, allowing retention by the membrane.
However, the agglomeration of this phase forms a layer on the
membrane surface that provides an additional resistance to the
ux and thus a lower value of the stabilized permeate ux com-
pared to the corresponding rened oil.
Importantly, as shown in Table 3, the percentage of glycerol in
the permeate was the same, 12%, both for rened soybean oil and
rened canola oil. These results indicate that probably the se-
paration of glycerol by ultraltration depends more on the acidity
of the oil used in the reaction than on its origin. Furthermore, the
higher acidity of the crude canola oil provided the best retention
of the dispersed phase and, consequently, of glycerol when com-
pared to the degummed soybean oil.
Samples of the permeates obtained from the ultraltration of
mixtures produced from vegetable oils are shown in Fig. 5. As
these samples passed through the rening process that removed
substances providing color to the oil, the permeates obtained from
the mixtures produced from rened oils exhibited a lighter color
compared to permeates of degummed soybean and crude canola
oils. In addition, it is possible to identify the dispersed phase in the
permeate of rened oils, in which the separation was not efcient.
Degummed Refined soybean Crude canola Refined canola
soybean
Fig. 5. Permeate obtained in the ultraltration of mixtures of the vegetable oils
studied.
 M.C.S. Gomes et al. / Journal of Membrane Science 496 (2015) 242249 247

Table 4
160
Permeate ux and glycerol concentration in the permeate in the ultraltration of
the mixture produced from crude canola oil.
140
Membrane (mm) Pressure (bar) Permeate ux (kg/h m2) Glycerol (%)
Permeate Flux (kg/h.m )

120
2

0.05 1.0 101.1 0.0137 0.003

100
80
60
40
20
0 increased ux with increasing pressure is due to driving force
0 10 20 30 40 50 60 70
Time (min)
Fig. 6. Permeate ux over time under different pressures in the ultraltration of
the mixture produced from crude canola oil. 0.05 mm membrane, T 50 C, 10%
acidied water.
3.4. Ultraltration of biodiesel produced from crude canola oil: in-
uence of membrane pore diameter and transmembrane pressure
In order to examine the inuence of acidity on the separation of
glycerol, ultraltration tests were performed with the mixture
produced from the transesterication of crude canola oil, whose
acidity is 2.4%, that is ten-fold higher than the acidity of the re-
ned canola oil. For the production of the reaction mixture,
transesterication was carried out at 30 C and 1:7.5, which was
the best condition when using an oil with high acidity [30].
The variation of permeate ux along time for the membranes of
0.05 mm and 20 kDa are presented in Figs. 6 and 7, respectively. For
all conditions evaluated, the concentration factor ranged from
1.3 to 1.4.
For both membranes at the three pressures examined, the
stabilized permeate uxes were close, ranging from 89 to 108
kg/h m2. Furthermore, the highest stabilized permeate ux was
obtained under the pressure of 2.0 bar. Meantime, higher rates of
ux decline were observed at this pressure, i.e. 24% for the 0.05 mm
membrane and 26% for the 20 kDa membrane. The increase in
pressure from 1.0 to 2.0 bar caused an increase in the permeate
ux. However, the pressure of 3.0 bar showed a lower stabilized
140
120
Permeate Flux (kg/h.m )
2
0.05 2.0 108.4 0.0127 0.005
0.05 3.0 97.8 0.0137 0.007
20 kDa 1.0 89.4 0.0117 0.008
20 kDa 2.0 96.4 0.0117 0.004
1.0 bar 20 kDa 3.0 90.5 0.0137 0.008
2.0 bar
3.0 bar
permeate ux in relation to the pressure of 2.0 bar.
In accordance with Chakrabarty et al. [40] and Yi et al. [41], the
80 90 100
across the membrane. Nevertheless, this increase is not linear,
since the appearance of additional resistances, such as the com-
paction of the phase retained on the membrane surface, which
reduce the ux.
The ultraltration of the crude canola oil, with membranes of
0.05 mm and 20 kDa, was effective in retaining glycerol, once the
values of glycerol in the permeate varied from 0.011% to 0.013%
under the conditions evaluated (Table 4). Besides that, the ux
curves versus time showed a high initial ux, which remained
almost constant throughout the operation, that is, the decline in
ux over time was small and ranged from 14% to 26%. These results
indicate that the retained phase did not cause membrane fouling.
Probably, this reduction of fouling is due to the larger size of
glycerol agglomerates, which are formed in the presence of free
fatty acids in the oil used. This retention of the dispersed phase
was conrmed by glycerol analysis (Table 4), since its highest
concentration in the permeate was 0.013%.
Comparing the results of ultraltration of crude canola bio-
diesel with the results of degummed soybean biodiesel presented
in a previous study [31], it was observed that the largest amount of
free fatty acids in the crude canola oil favors the formation of the
dispersed phase containing glycerol and thus the removal by the
membrane.
Yi et al. [41] investigated the inuence of acidity on the ultra-
ltration of oil-in-water emulsions and concluded that the clusters
of oil retained when used an acidic solution had a larger size
compared to ultraltration of an alkaline solution. The authors
explained that under acidic conditions, there is a reduction of the
electrostatic repulsive forces between the clusters, causing them
to bind to each other, forming larger agglomerates and thus re-
ducing membrane fouling.
Similar results of fouling reduction, given the formation of
larger clusters, were also obtained by Hesampour et al. [42] when
treated oily emulsions by ultraltration.

100 Also, this formation of larger clusters when used the mixture
produced from crude canola oil provided not only a greater re-
80 tention of glycerol by the membrane, but also the stabilization of
uxes at higher values (Fig. 7 and Table 4).
60 Therefore, regarding the lowest rate of ux decline over time
1.0 bar
and the quality of permeate achieved, the pressure of 1.0 bar is the
2.0 bar
most recommended for separation of glycerol and biodiesel pro-
40 3.0 bar duced by ethyl transesterication of the crude canola oil through
ultraltration. Wang et al. [25] employed ultraltration with
20
0.1 mm ceramic membrane for biodiesel purication after separa-
tion of the phases by settling, and concluded that the smaller re-
0 duction of ux along time (16.7%) was obtained at a pressure of 1
0 10 20 30 40 50 60 70 80 90 100
5 bar with a stabilized permeate ux of 300 L/h m2.
Time (min)
Fig. 7. Permeate ux over time under different pressures in the ultraltration of
3.5. Membrane cleaning
the mixture produced from crude canola oil. 20 kDa membrane, T 50 C, 10%
acidied water. Before starting the experiments, the membranes were
248 M.C.S. Gomes et al. / Journal of Membrane Science 496 (2015) 242249

characterized with deionized water at pressures of 1.0, 2.0 and treatment to reduce its acidity to around 0.8%, which was the
3.0 bar. The hydraulic permeability values of the membranes of acidity of the degummed soybean oil. As discussed earlier, it is
0.05 mm and 20 kDa were respectively 893 and 474 kg/h.m2 bar. important to have a certain acidity in the oil so that the metho-
The average values of permeate ux at 25 C and 1.0 bar used to dology developed in this work is adequate to separate glycerol
assess the cleaning of the membranes after each cleaning cycle, for using membranes.
the membranes of 0.05 mm and 20 kDa, were 867 and 450 kg/h m2, The caloric value of all biodiesel samples was above the
respectively. minimum required by the specication, 35,000 kJ/kg. The average
The methodology for membrane cleaning was efcient, and caloric values of the vegetable oil and of the biodiesel produced
after each washing cycle, the initial ux of the membranes was from it were similar. The caloric value of degummed soybean oil
restored with deionized water. Additionally, the cleaning in ul- was 39,230 kJ/kg and of the degummed soybean biodiesel was
trasonic bath reduced the water consumption compared to 38,154 kJ/kg. The caloric value of crude canola oil was 39,381 kJ/
washing the membrane in the module. Considering a complete kg and increased to 39,575 kJ/kg in the crude canola biodiesel.
washing cycle, corresponding to 6 washes with sodium hydroxide These values, as well as variations between oil and biodiesel were
solution, 6 washes with citric acid and 6 rinses with deionized similar to values reported by Lee et al. [7] in the transesterication
water, the total volume of water used is approximately 4.5 l. If the of soybean and canola oils.
same washing was performed in the micro-and ultraltration The ester content of both soybean biodiesel and canola bio-
module, the minimum water consumption would be about 15 l, diesel was higher than 97.0% in all samples, indicating that the
considering an alkaline wash, an acid wash and three rinses with reaction conditions used were appropriate for the transesterica-
deionized water, since the volume necessary for the recirculation tion of each oil. In other words, the permeates showed mass
in the module is 3 l. percentage of esters greater than the minimum required for
marketing as biodiesel, 96.5%.
3.6. Characteristics of the biodiesel produced Likewise, in all tests with the reaction mixture directly ultra-
ltered, the percentage of glycerol was below the value specied
The permeate of the mixture obtained from crude canola oil by the ANP for free glycerol, which is 0.02% in mass. Compared to
using the 20 kDa membrane, at a pressure of 1.0 bar and 10% the works in the literature using membrane technology in the
acidied water was subjected to evaporation under reduced purication of biodiesel [18,2428], the methodology used herein
pressure in a rotary evaporator at 70 C and  600 mmHg for has the advantage of eliminating the step of separation by settling
30 min and then characterized for specic mass, kinematic visc- and provide a biodiesel with a very low content of glycerol after a
osity, ester content, acidity, moisture, free glycerol content and single processing step.
caloric value. These results were compared to those obtained Taking into account that steps usually employed of settling and
using degummed soybean oil [31] and are listed in Table 5 to- washing can be replaced by membrane ltration and that this
gether with the values specied for the B100 biodiesel in Brazil process has provided lower content of glycerol than those pre-
[12]. sented in the literature, this study demonstrates a great prospect
The permeates showed viscosities within the specied range. for improving the separation and purication of biodiesel.
The canola biodiesel had a higher average viscosity than the soy-
bean biodiesel, which is a characteristic related to the fatty acid
prole of the oil. 4. Conclusions
Although ultraltration has removed most of the moisture,
since the initial concentration of water was 9.1% and the permeate Ceramic ultraltration membranes are used for the purication
before being evaporated showed an average moisture content of of biodiesel produced from ethyl transesterication of vegetable
0.45%, the vacuum evaporation treatment was not sufcient to oils with different characteristics, and enable to evaluate the in-
reduce the moisture in the nal product to less than 0.05%. It is uence of the acidity of the oil used on the membrane separation
possible that using a lower evaporating pressure for more than step.
30 min is adequate to reach the specied moisture value. The The ultraltration separation of glycerol from the reaction
temperature should not exceed 70 C, given the possibility of mixture obtained from rened canola oil is not effective in any
polymerization of the esters. condition evaluated. The comparison between the mixtures ob-
In the case of canola biodiesel, the high acidity can be reduced tained by transesterication of degummed soybean oil, rened
if the crude oil, which had an acidity of 2.4%, receives a pre- soybean oil, crude canola oil and rened canola oil demonstrate
that the efciency of membrane separation of glycerol depends on
Table 5 the acidity present in the oil. The highest percentage of acid in the
Characteristics of the biodiesel produced from degummed soybean oil and crude oil contributes to the agglomeration of the glycerol after addition
canola oil using ultraltration to remove glycerol. of acidied water in the nal reaction mixture. Accordingly, the
ultraltration of reaction mixtures produced from degummed
Characteristic Unit Limit Soybean biodiesel Canola biodie-
(Permeate) sel (Permeate)
soybean oil and crude canola oil is efcient in the removal of
GOMES et al. [31] glycerol. On the other hand, when used rened oils, the agglom-
erates in the emulsion show a diameter smaller than the mem-
Specic mass at kg/m3 850900 874 866 brane pore size, so that the glycerol permeates through the
20 C
2 membrane.
Kinematic visc- mm /s 3.06.0 4.38 4.85
osity at 40 C The efciency of the ultraltration of the mixture obtained
Ester content, % massa 96.5 97.2 97.5 from crude canola oil, which has the highest acidity, indicates that
mn the formation of dispersed phase clusters not only causes a greater
Free glycerol, % massa 0.02 0.014 0.011 retention of glycerol by the membrane, but also the stabilization of
max
Moisture, max mg/kg 500 976 1670
uxes at higher rates.
Acidity, max mg KOH/g 0.5 0.48 1.26 This process of biodiesel purication by ultraltration has the
Caloric value kJ/kg 35000 38154 39575 advantage of eliminating the step of separation by settling and
results in a biodiesel with very low content of glycerol after a
 M.C.S. Gomes et al. / Journal of Membrane Science 496 (2015) 242249
single processing step, and reduce the amount of water used for
washing.
Biodiesel produced using the methodology developed, both
from degummed soybean oil and crude canola oil, showed values
of ester content, glycerol content, viscosity, density and caloric
value within the range specied by the ANP for marketing.
References
[1] A. Demirbas, Biodiesel: A realistic fuel alternative for diesel engines, Springer
Verlag London Limited, London, 2008.
[2] F. Ma, M.A. Hannna, Biodiesel production: a review, Bioresour. Technol. 70
(1999) 115.
[3] J. Van Gerpen, Biodiesel processing and production, Fuel Process. Technol. 86
(2005) 10971107.
[4] G. Knothe, J. Van Gerpen, J. Krahl, L.P. Ramos, Manual de biodiesel, Edgard
Blcher, So Paulo (2006), p. 340 (Ed.).
[5] G.P.A.G. Pousa, A.L.F. Santos, P.A.Z. Suarez, History and policy of biodiesel in
Brazil, Energy Policy 35 (2007) 53935398.
[6] ANP, Agncia Nacional do Petrleo, Gs Natural e Biocombustveis. Relatrio
Mensal do Biodiesel, 2014. Available at http://www.anp.gov.br/?
pg 58819&m &t1 &t2 &t3 &t4 &ar &ps &cachebust
1325941677039, in April 2015.
[7] S.B. Lee, K.H. Han, J.D. Lee, I.K. Hong, Optimum process and energy density
analysis of canola oil biodiesel synthesis, J. Ind. Eng. Chem. 16 (2010)
10061010.
[8] EMBRAPA, Denio e histrico da canola, 2011. Available at http://hotsites.
sct.embrapa.br/diacampo/programacao/2008/potencialidades-da-canola-na-
producao-de-biodiesel, in April 2015.
[9] M.R. Monteiro, A.R.P. Ambrozin, L.M. Lio, A.G. Ferreira, Critical review on
analytical methods for biodiesel characterization, Talanta 77 (2008) 593605.
[10] H. Joshi, J. Toler, B.R. Moser, T. Walker, Biodiesel from canola oil using a 1:1 M
mixture of methanol and ethanol, Eur. J. Lipid Sci. Technol. 111 (2009)
464473.
[11] J. Van Gerpen, B. Shanks, R. Pruszko, D. Clements, G. Knothe, Biodiesel Pro-
duction Technology, Midwest Research Institute, National Renewable Energy
Laboratory, Colorado, United States (2004), p. 105.
[12] ANP, National Petroleum Agency. ANP Resolution Number 7, 19.03.08. Avail-
able at http://www.anp.gov.br/petro/biodiesel.asp in April 2015.
[13] M. etinkaya, F. Karaosmanoglu, Optimization of base-catalyzed transester-
ication reaction of used cooking oil, Energy Fuels 18 (2004) 18881895.
[14] I.M. Atadashi, M.K. Aroua, A.R.A. Aziz, N.M.N. Sulaiman, Renig technologies
for the purication of crude biodiesel, Appl. Energy 88 (2011) 42394251.
[15] L.C. Meher, V.S.S. Dharmagadda, S.N. Naik, Optimization of alkali-catalyzed
transesterication of Pongamia pinnata oil for production of biodiesel, Bior-
esour. Technol. 97 (2006) 13921397.
[16] F. Karaosmanoglu, B. Cigizoglu, M. Tter, S. Ertekin, Investigation of the re-
ning step of biodiesel production, Energy Fuels 10 (1996) 890895.
[17] J.H. Van Gerpen, E.G. Hammond, L.A. Johnson, S.J. Marley, L. Yu, I. Lee,
A. Monyem, Determining the inuence of contaminants on biodiesel proper-
ties, Iowa State University, United States (1996), p. 11 (The Iowa Soybean
Promotion Board).
[18] J. Saleh, A.Y. Tremblay, M.A. Dub, Glycerol removal from biodiesel using
membrane separation technology, Fuel 89 (2010) 22602266.
[19] L. Bournay, D. Casanave, B. Delfort, G. Hillion, J.A. Chodorge, New hetero-
geneous process for biodiesel production: A way to improve the quality and
the value of the crude glycerin produced by biodiesel plants, Catal. Today 106
(2005) 190192.
[20] I.M. Atadashi, M.K. Aroua, A.A. Aziz, Biodiesel separation and purication: a
review, Renew. Energy 36 (2011) 437443.
249
[21] M.A. Dub, A.Y. Tremblay, J. Liu, Biodiesel production using a membrane re-
actor, Bioresour. Technol. 98 (2007) 639647.
[22] P. Cao, A.Y. Tremblay, M.A. Dub, K. Morse, Effect of membrane pore size on
the performance of a membrane reactor for biodiesel production, Ind. Eng.
Chem. Res. 46 (2007) 5258.
[23] L. Cheng, Y. Cheng, S. Yen, J. Chen, Ultraltration of triglyceride from biodiesel
using the phase diagram of oil-FAME-MeOH, J. Membr. Sci. 330 (2009)
156165.
[24] H.Y. He, X. Guo, S.L. Zhu, Comparison of membrane extraction with traditional
extraction methods for biodiesel production, J. Am. Oil Chem. Soc. 83 (2006)
457460.
[25] Y. Wang, X. Wang, Y. Liu, S. Ou, Y. Tan, S. Tang, Rening of biodiesel by ceramic
membrane separation, Fuel Process. Technol. 90 (2009) 422427.
[26] J. Saleh, M.A. Dub, A.Y. Tremblay, Effect of soap, methanol, and water on
glycerol particle size in biodiesel purication, Energy Fuel 24 (2010)
61796186.
[27] J. Saleh, M.A. Dub, A.Y. Tremblay, Separation of glycerol from FAME using
ceramic membranes, Fuel Process. Technol. 92 (2011) 13051310.
[28] M.J. Alves, S.M. Nascimento, I.G. Pereira, M.I. Martins, V.L. Cardoso, M. Reis,
Biodiesel purication using micro and ultraltration membranes, Renew.
Energy 58 (2013) 1520.
[29] M.C.S. Gomes, N.C. Pereira, S.T.D. Barros, Separation of biodiesel and glycerol
using ceramic membranes, J. Membr. Sci. 352 (2010) 271276.
[30] M.C.S. Gomes, P.A. Arroyo, N.C. Pereira, Biodiesel production from degummed
soybean oil and glycerol removal using ceramic membrane, J. Membr. Sci. 378
(2011) 453461.
[31] M.C.S. Gomes, P.A. Arroyo, N.C. Pereira, Inuence of acidied water addition on
the biodiesel and glycerol separation through membrane technology, J.
Membr. Sci. 431 (2013) 2836.
[32] M.L. Pisarello, B.O. Dalla Costa, N.S. Veizaga, C.A. Querini, Volumetric method
for free and total glycerin determination in biodiesel, Ind. Eng. Chem. Res. 49
(2010) 89358941.
[33] D. Naviglio, R. Romano, F. Pizzolongo, A. Santini, A. De Vito, L. Schiavo, G. Nota,
S.S. Musso, Rapid determination of esteried glycerol and glycerides in tri-
glyceride fats and oils by means of periodate method after transesterication,
Food Chem. 102 (2007) 399405.
[34] S. Popovic, M. Djuric, S. Milanovic, M.N. Tekic, N. Lukic, Application of an ul-
trasound eld in chemical cleaning of ceramic tubular membrane fouled with
whey proteins, J. Food Eng. 101 (2010) 296302.
[35] M.O. Lamminem, H.W. Walker, L.K. Weavers, Mechanisms and factors inu-
encing the ultrasound cleaning of particle-fouled ceramic membranes, J.
Membr. Sci. 237 (2004) 213223.
[36] A. Al-Amoudi, R.W. Lovitt, Fouling strategies and the cleaning system of NF
membranes and factors affecting cleaning efciency, J. Membr. Sci. 303 (2007)
428.
[37] S.P. Sing, D. Sing, Biodiesel production through the use of different sources and
characterization of oils and their esters as the substitute of diesel: a review,
Renew. Sustain. Energy Rev. 14 (2010) 200216.
[38] B. Freedman, E.H. Pryde, T.L. Mounts, Variables affecting the yields of fatty
esters from transesteried vegetable oils, J. Am. Oil Chem. Soc. 61 (1984)
16381643.
[39] A. Murugesan, C. Umarani, T.R. Chinnusamy, M. Krishnan, R. Subramaniam,
N. Neduzchezhain, Production and analysis of biodiesel from non edible oils
a review, Renew. Sustain. Energy Rev. 13 (2009) 825834.
[40] B. Chakrabarty, A.K. Ghoshal, M.K. Purkait, Ultraltration of stable oil-in-water
emulsion by polysulfone membrane, J. Membr. Sci. 325 (2008) 427437.
[41] X.S. Yi, S.L. Yu, W.X. Shi, N. Sun, L.M. Jin, S. Wang, B. Zhang, C. Ma, L.P. Sun, The
inuence of important factors on ultraltration of oil/water emulsion using
PVDF membrane modied by nano-sized TiO2/Al2O3, Desalination 281 (2011)
179184.
[42] M. Hesampour, A. Kryzaniak, M. Nystrom, The inuence of different factors on
the stability and ultraltration of emulsied oil in water, J. Membr. Sci. 325
(2008) 199208.