You are on page 1of 7

Bioresource Technology 102 (2011) 1059–1065

Contents lists available at ScienceDirect

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

Influence of vegetable oils fatty-acid composition on biodiesel optimization


S. Pinzi a, J.M. Mata-Granados b, F.J. Lopez-Gimenez c, M.D. Luque de Castro b, M.P. Dorado a,⇑
a
Department of Chemical Physics and Applied Thermodynamics, EPS, Edificio Leonardo da Vinci, Campus de Rabanales, Universidad de Cordoba, 14071 Cordoba, Spain
b
Department of Analytical Chemistry, Edificio Marie Curie, Campus de Rabanales, Universidad de Cordoba, 14071 Cordoba, Spain
c
Department of Agricultural Engineering, ETSIAM, Edificio Leonardo da Vinci, Campus de Rabanales, Universidad de Cordoba, 14071 Cordoba, Spain

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

Article history: Biodiesel is an alternative fuel for diesel engines produced through transesterification of oleaginous feed-
Received 13 April 2010 stocks. To analyze the influence of the fatty-acid composition on biodiesel optimization, transesterifica-
Received in revised form 5 August 2010 tion of several vegetable oils has been studied. Reactions were carried out in flasks filled with vegetable
Accepted 18 August 2010
oils, heated to the reaction temperature and stirred at 1100 rpm. The reactions started when the meth-
Available online 24 August 2010
anol and potassium hydroxide solutions were added to the flasks. Concentration of catalyst, amount of
methanol, reaction temperature and time were optimized using a factorial design and a surface response
Keywords:
design. Also, a kinetics study was carried out to optimize the reaction time. Results showed that reaction
Kinetics study
Transesterification
parameters optimal values depend on the oil chemical and physical properties. It can be concluded from
Quality this field trial that the effect of both catalyst concentration and reaction time over the transesterification
Factorial design yield is greatly influenced by the saturation degree and fatty-acid chain length.
Surface response design Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction sel (Arzamendi et al., 2006; Dorado et al., 2004b; Freedman et al.,
1984; Vicente et al., 2005; Darnoko and Cheryan, 2000). All these as-
Biodiesel, consisting of mono-alkyl esters of vegetable oils or ani- pects are related to either biodiesel yield – or biodiesel purity
mal fats, is gaining interest and significance due to increased environ- (Vicente et al., 2007) – which is closely related to the fatty-acid
mental concerns and upsurge in petroleum prices (Knothe, 2006; methyl ester (FAME) concentration in the biodiesel phase.
Pinzi et al., 2009b). The most common way to produce biodiesel is Although optimization studies about many vegetable oils to
by transesterification. In this reaction, triglycerides (TG), the main produce biodiesel can be found in the literature, comparative stud-
component of vegetable oils, react with an alcohol to produce fatty- ies between them are missing (Pinzi et al., 2009b). In a recent
acid mono-alkyl esters and glycerol. Methanol is the most common study, Eevera et al. (2009) provided the optimum conditions of
alcohol used for transesterification and, in this case, the reaction is re- the transesterification reaction of some edible (coconut, palm,
ferred to as methanolysis. This reaction consists of three consecutive groundnut and rice bran) and some non-edible oils (pongamia,
reversible reactions with intermediate formation of diglycerides neem and cotton seed), applying a monofactorial study to each
(DG) and monoglycerides (MG). After reaction, glycerol is separated parameter to be optimized (catalyst concentration, temperature,
by either settling or centrifugation (Van Gerpen, 2005). methanol concentration and time), while the rest of the parame-
Biodiesel quality is extremely important to preserve the engine ters were kept constant.
from malfunction. Fuels with large amounts of mono-, di- and tri- Statistical optimization techniques (i.e. multifactorial design
glycerides are prone to coking and may cause the formation of and surface response methodology) are powerful tools that involve
deposits to injector nozzles, pistons and valves. Evidences of high the following general advantages: (1) more information per exper-
glycerides content in biodiesel are linked to increased values for vis- iment than unplanned approaches; (2) reduced number and cost of
cosity and carbon residue (Mittelbach and Remschmidt, 2004). Low experiments; (3) calculation of the interactions among experimen-
concentrations of mono-, di- and triglycerides can only be achieved tal factors within the range studied and with better process under-
by selecting optimum reaction conditions in order to produce max- standing; and (4) easier determination of the operational
imum methyl ester yield (Mittelbach and Remschmidt, 2004). conditions for scale-up the process (Vicente et al., 2007). In this
Research about transesterification of vegetable oils involves sense, these techniques have been used to develop and optimize
calculations of triglycerides conversion rate, changes in product biodiesel production from different raw materials (Vicente et al.,
composition during reaction and some quality parameters of biodie- 2007, 1998; Bouaid et al., 2007).
In this work, the synthesis of FAME from different feedstocks
⇑ Corresponding author. Tel.: +34 957 218332; fax: +34 957 218417. disclosing a wide range of fatty-acids composition and hence
E-mail address: pilar.dorado@uco.es (M.P. Dorado). iodine values (from 13.5 to 189, see Table 1) was studied. The

0960-8524/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biortech.2010.08.050
1060 S. Pinzi et al. / Bioresource Technology 102 (2011) 1059–1065

Table 1
Vegetable oils properties (MO, maize oil; SFO, sunflower oil; OOO, orujo olive oil; CO, coconut oil; LO, linseed oil; PO, palm oil).

Property MO SFO OOO CO LO PO


Kinematic viscosity (40oC) (mm2/s) 34.51 31.65 36.71 26.19 27.98 41.81
Free fatty acid content (%) 0.15 0.20 0.40 1.20 0.23 0.85
Iodine value (gI/100g) 118.1 134.7 99.8 33.5 189 58.32
Fatty-acid composition (%)
Caprylic (C8:0) 9.5
Decanoic (C10:0) 8
Lauric (C12:0) 41 0.5
Miristic (C14:0) <0.1 <0.10 18 1.5
Palmitic (C16:0) 13.00 6.90 11 9 5.5 45.5
Palmitoleic (C16:1) 0.8 0
Stearic (C18:0) 2 6 3 3.8 6 4
Oleic (C18:1) 33 26.5 75.2 7.5 21 38
Linoleic (C18:2) 50.8 66.5 9.5 2.7 14 10
Linolenic (C18:3) 1 0 0.5 0.5 53.5 0.5

process was developed and optimized by following a multifactorial Each sample was prepared in a 250-ml flask immersed in a water
design and a surface response methodology. A factorial design was bath equipped with a temperature controller (±0.2 °C of accuracy).
applied to determine the operation conditions affecting the transe- Agitation at 1100 rpm was provided by a magnetic stirrer. Up to
sterification process (reaction temperature, initial catalyst concen- 17 simultaneous reactions were conducted. Initially, each flask
tration, reaction time and methanol concentration), while the was filled with 100 g of vegetable oil and heated to the selected tem-
reaction yield was considered as the response variable. In addition, perature. Then, the methanol and potassium hydroxide solution was
the surface response methodology was used to calculate the opti- added to each flask-reactor. The reaction was assumed to start with
mum operation values that lead to the maximum biodiesel yield the stirring process (Dorado et al., 2004b).
for each selected feedstock. Reactions were stopped at different times cooling down in a
cold bath of ethylene glycolate at 10 °C. The samples were centri-
2. Methods fuged at 3500 rpm for 10 min to separate biodiesel from free glyc-
erol. Then, the mixture was washed with distilled water and
2.1. Vegetable oils centrifuged to remove the aqueous layer compounds by methanol,
residual catalyst and glycerol. The residual methanol and water
Biodiesel samples were produced after basic-catalyzed transe- were separated from biodiesel via rotary evaporation at 80 °C 1 h
sterification of six different vegetable oils showing a wide range (Ramos et al., 2009); then, the biodiesel produced from each sam-
of fatty-acid composition (Pinzi et al., 2009a). Sunflower oil ple was stored at 20 °C (Pinzi et al., 2009b).
(SFO), maize oil (MO) and orujo olive oil (OOO) were acquired from
KOIPESOL (Sevilla, Spain), linseed oil (LO) was purchased from 2.5. Analyses
Guinama (Valencia, Spain), palm oil (PO) and coconut oil (CO) were
acquired from Quimics Dalmau (Barcelona, Spain). The determination of FAME in SFME, MME, OOME and LME
samples was performed following the UNE EN 14103 standard.
2.2. Reagents However, due to their particular nature, a modified method of EN
14103 was needed for CME and PME samples (Schober et al.,
KOH and methanol were the catalyst and alcohol used to pro- 2006), in which modifications were: (1) a specific GC temperature
duce biodiesel, respectively. KOH pellets [85% p.a. CODEX program (initial oven temperature: 150 °C, increasing at 5 °C/min
(USP_NF)] and methanol ACS-ISO were acquired from PANREAC up to 220 °C, with 15-min constant final time) to make possible
(Barcelona, Spain). the chromatographic separation of short-chain fatty-acid esters
Heptadecanoic acid methyl ester from Fluka (Steinheim, Ger- (C8–C12) (Schober et al., 2006), and (2) an internal standard cali-
many) was used as internal standard for FAME determination, as bration method for short-chain fatty-acid esters (C8–C12) using
indicated in EN 14103 standard. For the determination of short- undecanoate methyl ester (C11:0), instead of the spiking standard
chain FAME (viz. coconut oil methyl esters), lauric and capric method proposed in the EN 14103 standard. This method makes
methyl esters standards were from Sigma–Aldrich (St. Louis, MO, possible to correct the unequal FID responses at all carbon atoms
USA) and undecanoate methyl ester was from Fluka (Steinheim, in the target methyl esters (Ackman and Sipos, 1964).
Germany) as internal standard.
2.6. Software
2.3. Instruments and apparatus
StatgraphicsÓ centurion XVI (StatPoint Technologies, Warren-
A Varian (Palo Alto, CA, USA) Star 3400 gas chromatograph ton-Virginia, USA) was used for building and analyzing the exper-
equipped with a flame ionization detector (FID) was used for indi- imental designs, and the response surface methodology (RSM) to
vidual separation and determination. A 60 m  0.25 mm VF-23 ms find out the optimum conditions and parameters involved in the
Varian capillary column (film thickness 0.25 lm) was used for the optimization.
determination of FAME.
3. Results and discussion
2.4. Transesterification process
3.1. Screening: linear step
Biodiesel was produced from sunflower oil, maize oil, orujo ol-
ive oil, linseed oil, palm oil and coconut oil. Key properties of these Prior to the synthesis of methyl esters by transesterification of
oils are shown in Table 1. SFO, OOO, MO, CO, PO and LO, a factorial design of experiments
S. Pinzi et al. / Bioresource Technology 102 (2011) 1059–1065 1061

was performed. A full 24 factorial design (including four factors, at from 0.8% to 1.2% (wt/wt), referred to mass oil, while the initial
two levels) allowing three degrees of freedom and involving 16 alcohol/oil molar ratio was in the range 4.2:1 to 5.4:1. The temper-
randomized runs plus three central points was applied to optimize ature was selected according to oil properties and taking into ac-
the synthesis of methyl esters by transesterification of SFO, OOO, count information obtained from previous works (Allen et al.,
MO, CO, PO and LO. The design was used to screen the most impor- 1999; Darnoko and Cheryan, 2000; Dorado et al., 2004b, 2002).
tant factors involved in biodiesel transesterification (reaction time, For this reason, a different range of temperature (from 40 °C to
reaction temperature, initial catalyst concentration and initial 60 °C) was selected for CO and PO, while for the rest of the oils,
alcohol/oil molar ratio) and to determine potential correlation be- the range of this variable was 25–45 °C. The experimental matrix
tween factors. for the factorial design is shown in Table 2. @The experiments were
Selection of parameters range was carried out on the basis of carried out randomly.
previous results (Pinzi et al., 2008; Vicente et al., 1998; Dorado The influence and interactions of the lineal model parameters on
et al., 2004a, 2004b; Freedman et al., 1984), considering the exper- the screening process for the response (yield of ester) and the stan-
imental installation and working conditions limits for each chem- dard response of each effect are shown in Table 3. As can be seen, in
ical species. The catalyst (KOH) influence was studied in the range the case of SFME, parameters A (reaction temperature), B (reaction

Table 2
Factorial designs for optimization of biodiesel production (MME, maize oil methyl ester; SFME, sunflower oil methyl ester; OOME, orujo olive oil methyl ester; CME, coconut oil
methyl ester; LME, linseed oil methyl ester, PME, palm oil methyl ester).

Range of factors used


Temperature (°C) Catalyst (wt%) Methanol/oil (molar ratio) Time (min)
Coded level 1 0 +1 1 0 +1 1 0 +1 1 0 +1
MME 25 35 45 0.8 1 1.2 4.2 4.8 5.4 10 25 40
SFME 25 35 45 0.8 1 1.2 4.2 4.8 5.4 10 25 40
OOME 25 35 45 0.8 1 1.2 4.2 4.8 5.4 10 25 40
CME 40 50 60 0.8 1 1.2 4.2 4.8 5.4 10 25 40
LME 25 35 45 0.8 1 1.2 4.2 4.8 5.4 10 25 40
PME 40 50 60 0.8 1 1.2 4.2 4.8 5.4 10 25 40
Type of experiment Run number T t Catalyst Methanol to oil molar ratio Yield (wt%)
MME SFME OOME CME LME PME
Linear step
1 +1 +1 1 +1 88.30 98.13 88.80 75.98 92.96 92.99
2 1 +1 +1 +1 90.50 94.88 86.78 76.28 90.75 94.92
3 +1 1 1 1 77.96 82.09 82.47 56.88 80.69 82.23
4 1 1 +1 +1 85.59 86.91 87.06 72.59 84.04 88.16
5 +1 +1 +1 +1 97.14 97.56 96.92 78.40 93.23 95.43
6 +1 1 +1 1 90.64 91.48 86.71 66.78 87.28 90.84
7 +1 +1 +1 1 89.86 92.96 90.19 72.20 89.60 90.26
8 1 +1 1 1 82.35 84.53 84.95 63.16 78.47 81.43
9 1 1 1 +1 80.31 84.93 86.54 62.17 78.22 78.91
10 1 +1 1 +1 85.51 90.05 86.13 69.73 84.19 82.87
11 1 1 1 1 76.21 78.64 81.15 61.17 75.12 75.94
12 +1 +1 1 1 86.86 90.35 84.73 64.75 84.21 87.97
13 1 1 +1 1 86.35 89.93 88.26 64.12 84.57 84.51
14 +1 1 1 +1 84.58 90.93 85.30 67.02 83.47 80.34
15 1 +1 +1 1 89.37 95.74 88.76 69.92 80.34 86.75
16 +1 1 +1 +1 90.51 96.94 89.81 74.41 88.99 90.41
Central points
17 0 0 0 0 91.40 91.84 89.38 72.61 87.00 88.18
18 0 0 0 0 91.96 90.41 89.43 71.06 87.42 88.90
19 0 0 0 0 91.07 91.96 89.52 73.26 87.12 87.65

Table 3
Influence of the lineal model parameters involved in the first design (screening process) on variable response (FAME yield). In bold, type coefficient of each significant factor
(factors were considered significant when the confidence level was over 95%) in the linear model for predicting yield of FAME using different raw materials (MME, maize oil
methyl ester; SFME, sunflower oil methyl ester; OOME, orujo olive oil methyl ester; CME, coconut oil methyl ester; LME, linseed oil methyl ester; PME, palm oil methyl ester).

Estimated effect MME SFME OOME CME LME PME


A: Reaction temperature (°C) 3.7075 ± 1.67659 5.22875 ± 0.740696 1.9125 ± 1.05033 1.66 ± 1.6936 5.59125 ± 0.98296 4.3725 ± 0.29053
B: Reaction time (min) 4.7175 ± 1.67659 5.16875 ± 0.740696 2.495 ± 1.05033 5.16 ± 1.6936 3.92125 ± 0.98296 4.66 ± 0.29053
C: Catalyst concentration (wt%) 7.235 ± 1.67659 5.71875 ± 0.740696 4.3025 ± 1.05033 6.23 ± 1.6936 5.18375 ± 0.98296 6.825 ± 0.29053
D: Methanol/oil (molar ratio) 2.855 ± 1.67659 3.45125* ± 0.740696 2.515 ± 1.05033 6.7 ± 1.6936 4.44625 ± 0.98296 2.5125 ± 0.29053
AB 0.1 ± 1.67659 0.02875 ± 0.740696 1.5925 ± 1.05033 0.4 ± 1.6936 0.927125 ± 0.98296 0.2975 ± 0.29053
AC 0.3775 ± 1.67659 0.60875 ± 0.740696 1.28 ± 1.05033 0.44 ± 1.6936 0.74125 ± 0.98296 2.3 ± 0.29053
AD 0.9475 ± 1.67659 1.46875 ± 0.740696 1.66 ± 1.05033 1.1 ± 1.6936 0.22875 ± 0.98296 1.2 ± 0.29053
BC 1.2725 ± 1.67659 1.44875 ± 0.740696 0.2075 ± 1.05033 1.435 ± 1.6936 1.66125 ± 0.98296 1.25 ± 0.29053
BD 0.3975 ± 1.67659 0.94125 ± 0.740696 0.015 ± 1.05033 0.11 ± 1.6936 2.68125 ± 0.98296 1.4375 ± 0.29053
CD 0.975 ± 1.67659 3.65625* ± 0.740696 0.8525 ± 1.05033 0.55 ± 1.6936 0.64125 ± 0.98296 0.6275 ± 0.29053

Standard errors are based on total error with 8 d.f.


1062 S. Pinzi et al. / Bioresource Technology 102 (2011) 1059–1065

time), C (catalyst concentration) and D (methanol/oil molar ratio) Table 5


have a significant and positive effect on the response variable, but Optimization steps of biodiesel production from different raw materials (MME, maize
oil methyl ester; SFME, sunflower oil methyl ester; OOME, orujo olive oil methyl
the interaction between methanol/oil molar ratio and concentration ester; CME, coconut oil methyl ester; LME, linseed oil methyl ester; PME, palm oil
of catalyst (effect CD in Table 3) results in a negative effect. To avoid methyl ester).
negative interactions between two factors (i.e. concentration of cat-
Factor Tested range Optimum
alyst and methanol/oil molar ratio), in the second experimental de- value
sign only concentration of catalyst was included, because it exhibits I design II design
a more significant effect as compared to methanol. (factorial (surface
The analysis of variance, estimated effects and interactions are design) response)
depicted in Table 4. As observed in the analysis of variance, and MME Yield: 98.67(wt%)
in agreement with other works (Bouaid et al., 2007; Dorado Reaction temperature (°C) 25–45 45–60 47.53
Catalyst concentration 0.8–1.2 1.2–1.8 1.92
et al., 2004b), the concentration of catalyst is the most significant
(wt%)
factor that influences biodiesel yield in every tested feedstock Methanol/oil (molar ratio) 4.2–5.4 5.4 5.4
(showing 99.7%, 99.9%, 99.6%, 99.3%, 99.8%, 99.9% of confidence le- Reaction time (min) 10–40 40 a

vel for MO, SFO, OOO, CO, PO and LO, respectively). Reaction time is SFME Yield: 99.70 (wt%)
other factor that significantly affects the yield of methyl ester from Reaction temperature (°C) 25–45 45–60 59.82
each tested vegetable oil. Catalyst concentration 0.8–1.2 1.1–2.1 1.81
The parameters that resulted in a significant effect over the (wt%)
Methanol/oil (molar ratio) 4.2–5.4 5.4 5.4
yield of the reaction (in bold, Tables 3 and 4), with the exception b
Reaction time (min) 10–40 40
of factor B (time of reaction) were included in the optimization
OOME Yield: 98.02 (wt%)
step, considering a wider range of values to achieve the highest Reaction temperature (°C) 25–45 45 45
yield of FAME for each feedstock. To study in depth the evolution Catalyst concentration 0.8–1.2 1.03–2.1 1.6
of transesterification over the time, a specific kinetics study was (wt%)
carried out after the optimization process. Methanol/oil molar ratio 4.2–5.4 5.4–6.6 6.03
a
Reaction time (min) 10–40 40
CME Yield: 90.01 (wt%)
3.2. Optimization step: Surface response Reaction temperature (°C) 40–60 60 60
Catalyst concentration 0.8–1.2 1.034–2.1656 1.7
(wt%)
Considering the conclusion drawn from Section 3.1, the inactive
Methanol/oil (molar ratio) 4.2–5.4 5.4–6.6 6.6
factors obtained for each raw material were kept constant, as Reaction time (min) 10–40 40 a

shown in Table 5. The effect of the parameters that significantly af-


LME Yield: 97.71 (wt%)
fect the yield of FAME was studied using surface response method- Reaction temperature (°C) 25–45 45–60 53
ology (SRM), which allows a quadratic model between response Catalyst concentration 0.8–1.2 1.034–2.1656 1.8
variable and factors, and optimization of the factors as a result (wt%)
(Massart et al., 1997). Table 5 depicts the optimal values of the fac- Methanol/oil (molar ratio) 4.2–5.4 5.4–6.6 6.02
a
Reaction time (min) 10–40 40
tors provided by each optimization step. The optimized values of
the surface response were calculated as maximum points of the PME Yield: 98.91(wt%)
Reaction temperature (°C) 45–65 65 65
second-order function studied in each model.
Catalyst concentration 0.8–1.2 1.034–2.1656 1.81
Fig. 1 shows the tridimensional response graphs of biodiesel (wt%)
yield predicted for the selected range of either catalyst concentra- Methanol/oil (molar ratio) 4.2–5.4 5.4–6.6 6.15
a
tion and reaction temperature or catalyst concentration and meth- Reaction time (min) 10–40 40
anol/oil molar ratio, depending on the optimized raw material, in a
Parameter optimized in the kinetics study.
agreement with the results in Table 5. In the case of LME, the meth- b
This parameter will be optimized during the kinetic study.
anol/oil molar ratio was fixed at 6:1. High slope in the surface
graphs represents high sensitivity of the response variable to incre-
ments of the corresponding factor. vegetable oil, the P-values of lack of fit are beyond 0.05, thus show-
Table 6 reports the regression coefficients of the quadratic mod- ing that mathematical models seem to be adequate for the ob-
el and the P-values of lack of fit. It can be seen that for every tested served data at the 95% confidence level.

Table 4
Analysis of variance of the lineal model parameters involved in the screening process (MME, maize oil methyl ester; SFME, sunflower oil methyl ester; OOME, orujo olive oil
methyl ester; CME, coconut oil methyl ester; LME, linseed oil methyl ester; PME, palm oil methyl ester).

MME SFME OOME CME LME PME


F-ratio P-value F-ratio P-value F-ratio P-value F-ratio P-value F-ratio P-value F-ratio P-value
A: Reaction temperature (°C) 4.89 0.058 49.83 0.0001 3.32 0.1061 0.96 0.3557 32.36 0.0005 194.28 0.0051
B: Reaction time (min) 7.92 0.0227 48.70 0.0001 5.64 0.0449 9.28 0.0159 15.91 0.004 220.67 0.0045
C: Catalyst concentration (wt%) 18.62 0.0026 59.61 0.0001 16.78 0.0035 13.53 0.0062 27.81 0.0008 473.34 0.0021
D: Methanol/oil (molar ratio) 2.90 0.1270 21.71 0.0016a 5.73 0.0436 15.65 0.0042 20.46 0.0019 64.15 0.0152
AB 0.00 0.9539 0.00 0.9700 2.30 0.1679 0.06 0.8192 0.98 0.3520 0.90 0.4430
AC 0.05 0.8275 0.68 0.43 1.49 0.2577 0.07 0.8016 0.57 0.4724 53.76 0.0181
AD 0.32 0.5875 4.08 0.703 2.52 0.1510 0.42 0.5342 0.05 0.8218 17.04 0.0540
BC 0.58 0.4696 3.83 0.0862 0.04 0.8483 0.72 0.4214 2.86 0.1295 19.63 0.0512
BD 0.06 0.8185 1.61 0.2395 0.00 0.9890 0.00 0.9498 7.44 0.0259 11.00 0.0945
CD 0.34 0.5769 24.37 0.0011a 0.66 0.4405 0.10 0.7602 0.43 0.5325 16.06 0.0598
a
Factors that show an antagonistic effect. To avoid negative interactions between factors, only the concentration of catalyst was included in the second experimental
design
S. Pinzi et al. / Bioresource Technology 102 (2011) 1059–1065 1063

Fig. 1. Estimated response surfaces of biodiesel yield produced from different raw materials. (a) CME, coconut methyl ester; (b) LME, linseed methyl ester; (c) MME, maize
methyl ester; (d) PME, palm oil methyl ester; (e) OOME, orujo olive methyl ester; and (f) SFME, sunflower methyl ester.

In addition, the models obtained for each tested feedstock may OOO, PO and CO, a greater curvature effect on quadratic effect of
be plotted in two-dimensional graphs (three graphs, instead of catalyst concentration than MO, SO and LO can be observed.
two, are needed when three factors are optimized), as shown in Near to the optimal point, the effect of temperature on transe-
Fig. 2. These called contour-plots are useful to determine optimum sterification of SFME and MME produces a mild slope in the surface
values provided by areas of high yield, besides the inclusion of eco- graphs, meaning a low sensitivity of the reaction yield to the incre-
nomical factors as moderate reaction temperature, low quantity of ment of this factor. The response surface permits to observe the
methanol and small amounts of catalyst, sorted by importance interaction between factors and different trends of sensitivity,
(Vicente et al., 2007). For each tested vegetable oil, different influ- depending on the factor involved in the model. For example, for
ence of the transesterification parameters on the response variable CME optimization the effect of the increment of methanol content
was found. The optimal amount of catalyst concentration for the in the yield of transesterification is very small for low catalyst con-
transeterification reaction resulted slightly higher than that re- centration; besides, the higher the concentration of catalyst, the
ported in the literature (Eevera et al., 2009; Darnoko and Cheryan, higher the sensitivity of the response variable to small amounts
2000; Vicente et al., 1998). Excess catalyst causes more triglycer- of methanol content in the reaction. Similar trend occurs for LME
ides participation in the saponification reaction leading to a optimization; with high concentration of catalyst, the yield is more
marked reduction in the ester yield (Vicente et al., 2007). For sensitive to small increments of temperature than when the reac-
tion is carried out with small amounts of catalyst.
Table 6
Statistical parameters of the second design (surface response) to optimize biodiesel 3.3. Third step. Study of the influence of time of reaction
production from the tested raw materials (MME, maize oil methyl ester; SFME,
sunflower oil methyl ester; OOME, orujo olive oil methyl ester; CME, coconut oil Once reaction conditions of each tested vegetable oil were opti-
methyl ester; LME, linseed oil methyl ester; PME, palm oil methyl ester). mized, the effect of reaction time on FAME yield was analyzed.
Statistical MME SFME OOME CME LME PME During the transesterification reaction, samples were taken at 30
parameter s, 1, 2, 5, 10, 35, 60, 90 and 120 min. Fig. 3 shows the progress of
R2 of model 94.38 99.18 96.92 92.63 96.66 93.217 yield of FAME during transesterification, considering a 2-h reaction
Lack of fit (P- 0.29 0.26 0.20 0.10 0.13 0.0696 time and the optimum reaction conditions for each tested vegeta-
value)a ble oil. It may be seen that every transesterification reaction pre-
Standard error 0.562 0.412 0.291 0.498 0.36 0.445
sents similar performance curve, with the exception of CME that
a
If P-value for lack of fit is greater than or equal to 0.05, thus the model appears exhibits the lowest yield of FAME. Oils with longer fatty-acid
to be adequate for the observed data at the 95% confidence level chains (OOO, MO, LO and SFO) achieve the optimal yield of FAME
1064 S. Pinzi et al. / Bioresource Technology 102 (2011) 1059–1065

Fig. 2. Contours of estimated response surface of biodiesel yield from the tested raw materials. (a) CME, coconut methyl ester; (b) LME, linseed methyl ester; (c) MME, maize
methyl ester; (d) PME, palm oil methyl ester; (e) OOME, orujo olive methyl ester; and (f) SFME, sunflower methyl ester.

Fig. 3. Transesterification yield vs. reaction time of tested vegetable oils.

after 20-min reaction, whereas PO and CO show the best 4. Conclusions


performance after 40-min reaction. Once the optimal yields were
achieved according to the European standard EN 14214, higher val- The effect of catalyst concentration is influenced by fatty-acid
ues of reaction time only retrieved small yield increments. composition. Vegetable oils composed by unsaturated fatty acids
S. Pinzi et al. / Bioresource Technology 102 (2011) 1059–1065 1065

show a directly proportional dependence between the concentra- Dorado, M.P., De Almeida, J.A., Schellert, C., Ballesteros, E., Löhrlein, H.-P., Krause, R.,
2002. An alkali-catalyzed transesterification process for high free fatty acid
tion of catalyst and yield, up to a maximum. Whereas, considering
feedstocks. Transactions of ASAE 45, 525–529.
vegetable oils with (mono)saturated fatty acids, amounts of cata- Eevera, T., Rajendran, K., Saradha, S., 2009. Biodiesel production process
lyst greater than the optimal value lead to soap production. Also, optimization and characterization to assess the suitability of the product for
optimal concentration of catalyst resulted higher than usually re- varied environmental conditions. Renewable Energy 34, 762–765.
Freedman, B., Pryde, E.H., Mounts, T.L., 1984. Variables affecting the yields of fatty
ported in the literature. Fatty acids chain length also seems to esters from transesterified vegetable oils. Journal of the American Oil Chemists
influence biodiesel conversion. Oils with longer fatty-acid chains Society 61, 1638–1643.
need half of the reaction time requested by oils comprising shorter Knothe, G., 2006. Analyzing biodiesel: standards and other methods. Journal of the
American Oil Chemists Society 83, 823–833.
fatty-acid chains to achieve maximum yield. Massart, D.L., Vanderginste, B.G.M., Buydens, L.M.C., De Jong, S., Lewi, P.J., Smeyers-
Verbeke, J., 1997. Handbook of Chemometrics and Qualimetrics, Part A. Elsevier,
Acknowledgements Amsterdam.
Mittelbach, M., Remschmidt, C., 2004. Biodiesel: The Comprehensive Handbook.
Martin Mittelbach ed., Graz, Austria.
The authors wish to acknowledge the support by the Spanish Pinzi, S., Capote, F. P., Jimenez, J. R., Dorado, M. P., De Castro, M. D. L., 2008.
Ministry of Science and Education (ENE2007-65490/ALT and Optimization and automatic on-line monitoring of biodiesel production from
vegetable oils. Alternative Fuels. Maribor, Slovenia.
HI2008-0229) and Junta de Andalucia, Spain (TEP-4994). Pinzi, S., Capote, F.P., Jimenez, J.R., Dorado, M.P., De Castro, M.D.L., 2009a. Flow
injection analysis-based methodology for automatic on-line monitoring and
References quality control for biodiesel production. Bioresource Technology 100, 421–427.
Pinzi, S., Garcia, I.L., Lopez-Gimenez, F.J., Luque De Castro, M.D., Dorado, G., Dorado,
M.P., 2009b. The ideal vegetable oil-based biodiesel composition: a review of
Ackman, R.G., Sipos, J.C., 1964. Application of specific response factors in the gas-
social, economical and technical implications. Energy & Fuels 23, 2325–2341.
chromatographic analysis of methyl esters of fatty acids with flame ionization
Ramos, M.J., Fernández, C.M., Casas, A., Rodríguez, L., Pérez, Á., 2009. Influence of
detectors. Journal of the American Oil Chemists Society 41, 377–380.
fatty acid composition of raw materials on biodiesel properties. Bioresource
Allen, C.A.W., Watts, K.C., Ackman, R.G., Pegg, M.J., 1999. Predicting the viscosity of
Technology 100, 261–268.
biodiesel fuels from their fatty acid ester composition. Fuel 78, 1319–1326.
Schober, S., Seidl, I., Mittelbach, M., 2006. Ester content evaluation in biodiesel from
Arzamendi, G., Arguiñarena, E., Campo, I., Gandía, L.M., 2006. Monitoring of
animal fats and lauric oils. European Journal of Lipid Science and Technology
biodiesel production: simultaneous analysis of the transesterification products
108, 309–314.
using size-exclusion chromatography. Chemical Engineering Journal 122,
Van Gerpen, J., 2005. Biodiesel processing and production. Fuel Processing and
31–40.
Technology 86, 1097–1107.
Bouaid, A., Martinez, M., Aracil, J., 2007. A comparative study of the production of
Vicente, G., Coteron, A., Martinez, M., Aracil, J., 1998. Application of the factorial
ethyl esters from vegetable oils as a biodiesel fuel optimization by factorial
design of experiments and response surface methodology to optimize biodiesel
design. Chemical Engineering Journal 134, 93–99.
production. Industrial Crops and Products 8, 29–35.
Darnoko, D., Cheryan, M., 2000. Kinetics of palm oil transesterification in a batch
Vicente, G., Martinez, M., Aracil, J., 2007. Optimisation of integrated biodiesel
reactor. Journal of the American Oil Chemists Society 77, 1263–1267.
production. Part I. A study of the biodiesel purity and yield.. Bioresource
Dorado, M.P., Ballesteros, E., Lopez, F.J., Mittelbach, M., 2004a. Optimization of
Technology 98, 1724–1733.
alkali-catalyzed transesterification of Brassica carinata oil for biodiesel
Vicente, G., Martinez, M., Aracil, J., Esteban, A., 2005. Kinetics of sunflower oil
production. Energy & Fuels 18, 77–83.
methanolysis. Industrial & Engineering Chemistry Research 44, 5447–5454.
Dorado, M.P., Ballesteros, E., Mittelbach, M., Lopez, F.J., 2004b. Kinetic parameters
affecting the alkali-catalyzed transesterification process of used olive oil.
Energy & Fuels 18, 1457–1462.

You might also like