Optimization and kinetics of sunflower oil methanolysis catalyzed by calcium oxide-based catalyst derived from palm kernel shell biochar

Fuel 163 (2016) 304–313

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Optimization and kinetics of sunflower oil methanolysis catalyzed by calcium oxide-based catalyst derived from palm kernel shell biochar Milan D. Kostic´ a, Alireza Bazargan b, Olivera S. Stamenkovic´ a, Vlada B. Veljkovic´ a, Gordon McKay b,c,⇑ a

Faculty of Technology, University of Niš, Bulevar oslobodjenja 124, Leskovac 16000, Serbia Department of Chemical and Biomolecular Engineering, Hong Kong University of Science and Technology, Clearwater Bay, Hong Kong c Division of Sustainable Development, College of Science, Engineering and Technology, Hamad bin Khalifa University, Qatar Foundation, Doha, Qatar b

h i g h l i g h t s
CaO-rich catalyst obtained from palm kernel shell biochar (PKSB) has promising potential for biodiesel production.
Sunflower oil methanolysis catalyzed by PKSB-based catalyst is optimized.
Optimum reaction conditions ensure the best FAME content of 99%.
Reaction rate law is changing- and first-order with respect to TAG and FAME.
PKSB catalyst can be reused without any treatment in three consecutive cycles.

a r t i c l e

i n f o

Article history: Received 1 July 2015 Received in revised form 12 September 2015 Accepted 16 September 2015 Available online 26 September 2015 Keywords: Biodiesel catalyst Calcium oxide Methanolysis Kinetics Optimization Palm kernel shell biochar

a b s t r a c t Sunflower oil methanolysis over CaO-based palm kernel shell biochar (PKSB) catalyst was assessed by coupling full factorial design with modeling, optimization and kinetic studies. According to the analysis of variance, the effect of reaction temperature and methanol-to-oil molar ratio on the fatty acid methyl ester (FAME) synthesis is significant, while the effect of catalyst loading is statistically negligible. The optimum reaction conditions are found to be catalyst loading of 3 wt%, temperature of 65 °C and methanol-to-oil molar ratio of 9:1, ensuring the best FAME content of 99%. The kinetic model of the methanolysis of sunflower oil, catalyzed by PKSB-based catalyst, combines the changing- and firstorder reaction rate laws with respect to triacylglycerols and FAMEs, respectively. The high activation energy (108.8 kJ/mol) indicates the temperature sensitivity of the reaction. The CaO-based PKSB catalyst can be reused without any treatment in three consecutive cycles with no significant drop in activity. Since the calcium content in the biodiesel product is higher than the standard limit, the overall process should include a purification stage. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Biodiesel is an alternative fuel that is technically applicable, economically competitive and environmentally beneficial. Conventionally, biodiesel production is performed by homogeneous basecatalyzed transesterification (most frequently methanolysis) of plant oils and animal fats because of moderate reaction conditions and a short reaction time. However, these processes have several drawbacks such as the impossibility of catalyst reuse, difficulty of operation, energy demand for the separation and purification of biodiesel, and the production of a large amount of wastewater ⇑ Corresponding author at: Department of Chemical and Biomolecular Engineering, Hong Kong University of Science and Technology, Clearwater Bay, Hong Kong. Tel.: +852 2358 8412; fax: +852 2358 0054. E-mail address: [email protected] (G. McKay). http://dx.doi.org/10.1016/j.fuel.2015.09.042 0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.

in the purification stage [1]. These problems can be overcome using heterogeneous (solid) catalysts that offer many benefits such as easy separation from the reaction mixture, less risk of corrosion, less environmental hazard and possibility of reuse. Among the solid catalysts, calcium compounds in the form of neat, loaded, mixed or supported CaO, are very often used as catalysts for transesterification of oily feedstocks. The CaO-based catalysts possess high basicity, require mild reaction conditions, produce high biodiesel yield, have low or no cost and are easy to prepare from widely-available materials of natural origin or waste precursors. In the past few years, ashes of different origins have been tested as catalysts for biodiesel production [2–4]. Recently, Bazargan et al. [5] have shown that calcium carbonate contained in palm kernel shell biochar (PKSB) is a promising source material for producing a CaO-based catalyst with activity for the

M.D. Kostic´ et al. / Fuel 163 (2016) 304–313

305

Nomenclature Adeq.Prec. cA cA0 cB0 cC cC0

C.V. FAME k K K0

adequate precision concentration of TAG (mol/dm3) initial concentration of TAG (mol/dm3) initial concentration of methanol (mol/dm3) concentration of FAME (mol/dm3) model parameter, i.e. the hypothetical initial FAME concentration corresponding to the initial available active catalyst surface (mol/dm3) coefficient of variation (%) fatty acid methyl ester reaction rate constant (dm6/(mol2 min)) model parameter defining the TAG affinity for the catalyst active sites (mol/dm3) TAG affinity for the catalyst active sites (dm3/mol)

transesterification of sunflower oil. CaO-based catalysts are usually employed in powder form in batch reactors, although they are also used in packed-bed reactors [6,7]. For increasing the process efficiency, knowledge of the effect of reaction conditions on biodiesel yield and transesterification reaction rate are of great importance. Statistical [8–14] and experimental [15–18] optimizations have been used to evaluate the effect of reaction conditions on CaO-catalyzed transesterification reactions and to determine the optimal reaction conditions ensuring the highest biodiesel yield. Since the experimental approach is time consuming and cost-excessive, statistical methods are more powerful. So far, response surface methodology (RSM) combined with the central composite [8–10,14], Box–Behnken factorial [11,12] or factorial [13] design has been used. Various kinetic models supposing zero [19,20] or first order [21–28] reactions with respect to triacylglycerols (TAGs) have been used for methanolysis reactions over calcium-based catalysts. In fact, the reaction order varies during the transesterification reaction from the zeroth order to the first order [24,29]. More complex models [7,26] involve the liquid–liquid mass transfer limitation to predict the reaction rate during the whole course of sunflower oil methanolysis. These kinetic models have been recently verified for calcium-based catalysts including CaO [30,31]. In the present work, the CaO-rich PKSB catalyst was used in sunflower oil methanolysis under moderate reaction conditions and atmospheric pressure. The influence of the most important operational factors, namely reaction temperature, methanol-tooil molar ratio and catalyst loading, on the sunflower oil methanolysis over the PKSB catalyst were studied in the following ranges: 45–65 °C, 9:1–15:1 and 3–7% (based on the oil weight). The experimental data were generated by a 33 full factorial design of experiment (DOE) with five central points. This study included both the optimization and kinetics of the methanolysis reaction over the PKSB catalyst aimed at developing the statistical and kinetic models describing the relationship between fatty acid methyl ester (FAME) content, and the operating variables and the time variation of the TAG conversion degree, respectively. In addition, leaching of calcium and reusability of the PKSB catalyst was evaluated in order to estimate its potentials for biodiesel production at a commercial scale. Finally, the biodiesel product was characterized in accordance with biodiesel standard requirements. 2. Materials and methods 2.1. Materials The PKSB is in the form of a fine black powder (particle size ranged from about 1 lm to more than 100 lm). Obtained from a

km MRPD n R2 R2adj R2pred t T TAG xA ya ym yp

apparent reaction rate constant (min1) mean relative percent deviation (%) number of experimental runs coefficient of determination adjusted coefficient of determination predicted coefficient of determination time (min) temperature (K) (Eq. (9)) or (°C) (Eqs. (11) and (12)) Triacylglyceride degree of TAG conversion experimental values of the TAG conversion degree (%) mean value of the TAG conversion degree (%) predicted values of the TAG conversion degree (%)

gasifier for electricity production, it was calcined at 800 °C for 2 h under atmospheric pressure immediately before use as a catalyst [5]. The catalyst has been thoroughly characterized elsewhere, showing that after calcination, the catalyst is predominantly composed of calcium oxide. In fact, the calcination process breaks down the calcium carbonate content of the kernel shell biochar into gaseous carbon dioxide which leaves the system, and calcium oxide which remains. Also, the basic strength of the catalyst is in the range 11.0–15.0, with a total basicity of 0.516 mmol/g [5]. Edible sunflower oil (Dijamant, Zrenjanin, Serbia) was purchased in a local shopping store. The acid value of the oil was 0.2 mg KOH/g. Certified methanol of 99.5% purity was purchased from Zorka Pharma (Šabac, Serbia). Methanol, 2-propanol and n-hexane, all of HPLC grade, were purchased from LAB-SCAN (Dublin, Ireland). Hydrochloric acid (36 wt%), was purchased from Centrohem (Belgrade, Serbia). Anhydrous sodium carbonate and sodium sulfate were obtained from Sigma Aldrich (Saint Louis, USA). 2.2. Equipment and reaction conditions The methanolysis reaction of sunflower oil over the PKSB catalyst was performed at atmospheric pressure in a 250 mL threeneck glass flask, equipped with a reflux condenser and a magnetic stirrer. The flask was placed in a water tank keeping the temperature constant at the desired level by circulating water from a thermostated bath. A 33 full factorial design with five central points was used to optimize the reaction at methanol-to-oil molar ratios of 9:1, 12:1 and 15:1, PKSB catalyst amounts of 3%, 5% and 7% (based on the oil weight) and temperatures of 45, 55 and 65 °C. These reaction conditions were chosen by considering the literature data on the methanolysis reaction catalyzed by CaO [32–37]. So far, the CaO-catalyzed methanolysis of sunflower has been studied in the ranges of methanol-to-oil molar ratio and catalyst loading of 6:1 to 18:1 and 1% to 10% (based on the oil weight), respectively. Hence, we narrowed ranges of both methanol-to-oil molar ratio and catalyst loading to 9:1 to 15:1 and 3–7%, respectively. The upper temperature level (65 °C) is actually methanol’s boiling temperature, and the lower level (45 °C) ensures an acceptable reaction rate. The catalyst loading was varied around the reported optimum value for CaO-based catalysts (5%) [5]. Sunflower oil has been used as a TAG source in a few studies of the kinetics [28,29] and optimization [38] of the methanolysis reaction catalyzed by neat CaO and quicklime. The methanolysis reaction was also performed under the optimal reaction conditions determined by the applied design of experiments and RSM. The final reaction mixture was used to obtain biodiesel for characterization and to evaluate calcium leaching.

306

M.D. Kostic´ et al. / Fuel 163 (2016) 304–313

cA cA0

2.3. Methanolysis procedure

xA ¼ 1 

The desired amounts of methanol and the catalyst were added to the flask. The methanol-catalyst suspension was thermostated at the desired temperature while stirred at 900 rpm for 30 min. The stirrer was turned off, and the corresponding amount of sunflower oil, thermostated separately at the same temperature, was added to the reaction flask. Then, the stirrer was switched on again and the reaction was timed. During the reaction, the samples were taken from the reaction mixture and immediately quenched by adding a required amount of aqueous hydrochloric acid solution (5 M) to neutralize the catalyst. After centrifugation (Sigma 2-6E, Germany; 3500 rpm; about 700g) for 15 min, the supernatant was withdrawn, dissolved in a solution of 2-propanol and n-hexane (5:4 v/v) in a ratio of 1:200 and filtered through a 0.45 lm Millipore filter. The filtrate was used for HPLC analysis.

The calcium concentration in biodiesel, expressed as mg of Ca per kg (ppm), was determined by atomic absorption spectrometry, flame technique. For this purpose, the samples were prepared by microwave digestion (MBS-9, CEM Innovators, Great Britain) with a mixture of concentrated HCl and HNO3 acids (metal-free). After filtering, all samples were diluted with metal-free ultrapure water.

2.4. Biodiesel separation, purification and characterization The reaction mixture from the methanolysis conducted under the optimal reaction conditions was gravitationally separated into crude biodiesel (upper layer, supernatant) and the alcoholic phase containing glycerol and methanol (lower layer, subnatant/infranatant). The crude biodiesel obtained in the optimized experimental run was purified by wet washing using distilled water or acidified water for neutralizing crude biodiesel followed by water rinsing and the procedure of Alba Rubio et al. [39]. Water washing: Crude biodiesel was washed using distilled water (10% of biodiesel mass) at 25 °C. Also, crude biodiesel was first neutralized with the 5 M HCl solution and then washed with distilled water (10% of biodiesel mass, 25 °C). After separating the aqueous phase, the ester phase was dried using anhydrous sodium sulfate. In both cases, the emulsion was vigorously stirred (500 rpm). Method of Alba Rubio and coworkers: [39]: The ester phase was rinsed at first with methanol (50% based on the crude biodiesel mass) containing anhydrous sodium carbonate (5% based on the crude biodiesel mass) at the boiling temperature for 4 h under agitation by a magnetic stirrer (1000 rpm) and then, after gravitational separation and filtration, with distilled water (10% of biodiesel mass, 25 °C) under agitation (500 rpm) for 1 h. Water was gravitationally separated from the ester phase, which was then dried by adding anhydrous sodium sulfate. The purified biodiesel was characterized by the specified methods according to the EN 14214 standard for biodiesel. Its physical and chemical properties, namely density, kinematic viscosity, iodine value, acid value, water content, FAME content, as well as the contents of monoacylglycerols (MAGs), diacylglycerols (DAGs) and TAGs, were determined according to the appropriate standard methods. 2.5. Catalyst reusability test This test was conducted under the optimal reaction conditions. After the completion of the reaction, the catalyst was separated from the ester phase by centrifugation (3500 rpm, 15 min) and reused without any treatment (no regeneration or recalcination). 2.6. Analytical methods The composition of the reaction mixture was determined by the HPLC method described elsewhere [40]. The TAG conversion degree was calculated from the percentage of TAG in the ester/oil fraction of the reaction mixture at the beginning (cA0) and at any time (cA) of the reaction:

ð1Þ

2.7. Statistical and kinetic modeling 2.7.1. Statistical modeling and optimization A 33 full factorial design of experiments was applied to find out the influence of the operational factors on the FAME content achieved within 4 h. Factor levels (Table 1) were chosen by considering the operational limits of the experimental setup. The experimental matrix of the applied 33 factorial design with five central points along with the observed FAME contents are shown in Table S1 (Supplementary Material). The experimental data values on FAME content were analyzed using RSM and the second-order polynomial equation:

y ¼ bo þ b1 X 1 þ b2 X 2 þ b3 X 3 þ b12 X 1 X 2 þ b13 X 1 X 3 þ b23 X 2 X 3 þ b11 X 21 þ b22 X 22 þ b33 X 23

ð2Þ

where y represents FAME content (response), X1, X2 and X3 are reaction temperature, methanol-to-oil molar ratio and catalyst amount (independent variables), respectively; b0 is the intercept and bi, bii, bij (i = 1, 2, 3 and j > i) are linear, quadratic and interactive regression coefficients, respectively. Process factors and their actual (uncoded) and coded levels are given in Table 1. Analysis of variance (ANOVA) was used for evaluating the statistical significance of the process factors and the quality of the model fit. The optimal conditions for maximizing the FAME content in the ester–oil phase were determined by solving the model equation. 2.7.2. Kinetic modeling For modeling the kinetics of sunflower oil methanolysis over the PKSB catalyst, the three-parametric kinetic model of Miladinovic´ et al. [29] was adopted:

dxA ð1  xA Þ  ðcC0 þ 3  cA0  xA Þ ¼ km K þ cA0 ð1  xA Þ dt

ð3Þ

where xA is the conversion degree of TAG, t is time, km is the apparent reaction rate constant, cC0 is the hypothetical initial FAME concentration corresponding to the initial available active catalyst surface, cA0 is the initial TAG concentration, and K is the model parameter defining the TAG affinity for the catalyst active sites. The assumptions made in deriving the model can be found elsewhere [29]. This model has recently been proven for several calcium-based catalysts (neat CaO, quicklime, CaOZnO and calcium hydroxide) used in the transesterification of sunflower oil [30]. 2.7.3. Computer software The statistical analysis was performed using Design-Expert 7.0.0 (Stat-Ease Inc., Minneapolis, MN). Kinetic parameters of Eq. (3) were calculated using Polymath v.6.10 by minimizing the objective function by the Levenberg–Marquardt algorithm. Using the calculated values of the kinetic parameters, values of the TAG conversion at various reaction times were calculated by the same software. The significance of the model was statistically evaluated based on the mean relative percent deviation (MRPD) and the coefficient of determination (R2) calculated by Eqs (4) and (5), respectively:

M.D. Kostic´ et al. / Fuel 163 (2016) 304–313

307

Table 1 Process factors and their levels.

a

Symbol

Factor

Unit

Actual levels

X1 X2 X3

Temperature Methanol-to-oil molar ratio Catalyst amounta

°C mol/mol %

Coded levels

Low

Middle

High

Low

Middle

High

45 9 3

55 12 5

65 15 7

1 1 1

0 0 0

1 1 1

Based on the oil weight.

MRPD ¼

 n  y  ya;i  100 X  p;i  n i¼1  ya;i 

Pn R2 ¼ Pi¼1 n

ð4Þ

ðyp;i  ya;i Þ2

i¼1 ðyp;i

ð5Þ

 y m Þ2

where yp and ya are the predicted and experimental values of the TAG conversion (%), ym is the mean value of the TAG conversion (%), and n is the number of experimental runs. Other statistical criteria were also used, including: the p-value of the lack of fit, the adjusted coefficient of determination R2adj , the predicted coefficient of determination R2pred , the coefficient of variation C.V. and the adequate precision Adeq.Prec. 3. Results and discussion 3.1. Modeling and optimization of the methanolysis reaction 3.1.1. Analysis of variance (ANOVA) The operational variables that influence the FAME content were screened out by using ANOVA. The results of the ANOVA screening are summarized in Table 2 through the sum and mean of squares of residuals, degrees of freedom, as well as F- and p-values. In terms of the statistical significance of the overall model, all three factors, their squares and two-way interactions for FAME content were evaluated from their F- and p-values. Only the terms having the p-value less than 0.05 affect the FAME content significantly at the 95% confidence level. The model F- and p-values of 522.4 and <0.0001, respectively imply that the model is statistically significant. The lack of fit is not significant relative to the pure error since its p-value (0.096) is higher than 0.05, meaning that the model is adequate for predicting the FAME content within the applied ranges of reaction temperature, methanol-to-oil molar ratio and catalyst loading. Of the individual operational variables, the reaction temperature (X1) and the methanol-to-oil molar ratio (X2)

have a statistically significant effect on FAME content, while the effect of PKSB catalyst loading (X3) is statistically insignificant. The most important operational variable is reaction temperature having the highest F-value (3320.7), followed by the methanolto-oil molar ratio (F-value = 85.7). From the two-way interactions and squares of individual variables, only the interactions between the reaction temperature and methanol-to-oil molar ratio (X1  X2) and between the methanol-to-oil molar ratio and the catalyst loading (X2  X3) as well as the square of reaction temperature (X 21 ) significantly affect the FAME content. According to Table 2, the sums of the F-values for individual process factors (3407.5) and their two-way interactions (53.6) indicate that the effect of individual process factors is relatively more significant than that of their interactions. Physically, this means that the reaction temperature and methanol-to-oil molar ratio affect the FAME content more as independent variables, while the link between the mechanisms of their effects is less important. These results imply that by increasing both the reaction temperature and methanol concentration, FAME formation is favored and accelerated. In addition, the increase of the reaction temperature promotes TAG mass transfer toward active sites at the surface of the PKSB catalyst particles by reducing the viscosity of the liquid phases involved, thus enhancing the overall process rate. Methanol reacts with the active sites at the surface of the PKSB catalyst particles, forming methoxide ions that are the actual catalysts of the methanolysis reaction. Since the reaction at the surface of the catalyst particles is faster than the TAG mass transfer toward the PKSB catalyst particles, catalyst loading does not affect FAME formation. In other words, the number of active sites participating in the catalysis of the methanolysis reaction is effectively large enough at the lowest PKSB catalyst loading. The present result partly agrees with the previous studies on the influence of reaction conditions on FAME yield in heterogeneously catalyzed alcoholysis of different oils. Generally, reaction temperature, catalyst amount [8–13,41,42] and reaction time [8,11,13] have significant effects on FAME yield. Different observations

Table 2 ANOVA results.

a

Source

Sum of squares

Mean square

F-value

p-value Prob > F

Model X1 X2 X3 X1  X2 X1  X3 X2  X3 X 21

17,739.7 12,529.4 323.4 3.9 146.3 2.3 53.8 4084.0

9 1 1 1 1 1 1 1

1971.1 12,529.4 323.4 3.9 146.3 2.3 53.8 4084.0

522.40 3320.73 85.72 1.04 38.77 0.62 14.25 1082.39

<0.0001 <0.0001 <0.0001 0.319a <0.0001 0.439a 0.001 <0.0001

X 22

4.1

1

4.1

1.09

0.307a

X 23

8.8

1

8.8

2.33

0.141a

Residual Lack of fit Pure error Cor total

83.0 76.2 6.8 17,822.7

22 17 5 31

3.8 4.5 1.4

3.29

0.096a

Not significant at the 95% confidence level.

df

M.D. Kostic´ et al. / Fuel 163 (2016) 304–313

308

have been reported regarding the effect of alcohol-to-oil molar ratio. The insignificant effect of the methanol-to-oil molar ratio on ester yield was reported for the methanolysis of palm oil catalyzed by CaO/Al2O3 [14] and pork lard catalyzed by calcium manganese oxide [10]. The less significant effects of the methanol-to-oil molar ratio on the FAME content compared to the effect of catalyst amount and reaction time was observed in the continuous microwave-assisted Jatropha curcas oil methanolysis in the presence of KOH-impregnated CaO catalyst [11]. On the other hand, other researchers have observed significant influence of the molar ratio on the ester synthesis [9,12,41,42]. According to most studies, catalyst loading has a significant influence on FAME content, which disagrees with the present study. No influence of the PKSB loading higher than 5% on FAME synthesis was observed in our previous study of sunflower oil methanolysis performed at 60 °C and a methanol-to-oil molar ratio of 9:1 [5]. The same influence of catalyst loading has been observed with transesterification of various vegetable oils oil over quicklime [29] and supported CaO catalysts [18,43]. In some cases, even the decrease in the oil conversion is observed with increasing the catalyst amount above a certain level [44]. These negative effects of higher catalyst loading were possibly due to enhanced viscosity of the three-phase reaction mixture creating the mixing problem and increasing the mass transfer resistance in the system [43,44]. It has also been demonstrated that higher catalyst amounts can lead to the decrease of available catalytically active surface sites [28], probably due to the clustering of catalyst particles under the mentioned mixing problem. The independence of FAME formation from the quadratic term of catalyst loading as well as the interaction between reaction temperature and catalyst amount has also been previously observed [10,12]. 3.1.2. Prediction of FAME content by RSM model The experimental data on the FAME content achieved in the sunflower oil methanolysis over the PKSB catalyst were analyzed by the multiple nonlinear regression method to develop the second-order polynomial equation based on the coded and uncoded (actual) values of the three operational factors: Actual factors

y ¼ 867:551þ30:428X 1 þ8:076X 2 0:478X 3 0:116X 1 X 2  0:022X 1 X 3 þ0:353X 2 X 3 0:239X 21 0:084X 22 0:277X 23 ð6Þ Coded factors

y ¼ 96:923þ26:383X 1 þ4:239X 2 0:467X 3 3:492X 1 X 2  0:442X 1 X 3 þ2:117X 2 X 3 23:893X 21 0:760X 22 1:110X 23 ð7Þ which presents FAME content (y) as a function of reaction temperature (X1), methanol-to-oil molar ratio (X2) and catalyst loading (X3). Since the coded variables in the regression equation, Eq. (7), were in the normalized form, the coefficients gave direct measures of the contribution of each factor and their interactions. Reaction temperature has the largest coefficient (26.383), indicating that it is the most dominating factor that affects FAME content, followed by methanol-to-oil molar ratio (4.239). The statistical significance of the regression model is also judged based on several additional statistical criteria. First, the coefficient of determination (R2 = 0.995) indicates an excellent fit, meaning that the fitted model explained about 99.5% of the total variation in the FAME content. A good agreement among the coefficient of determination R2, the adjusted coefficient of determination (R2adj = 0.993) and the predicted coefficient of determination (R2pred = 0.991) advocates the high correlation between actual and

predicted FAME contents. In addition, the coefficient of variation (C.V. = 2.4%) indicates a high degree of precision and reliability of the developed models. If C.V. < 10%, then the model is considered reproducible [45]. The high value of Adeq.Prec. (60.8) indicates an adequate signal for the developed model, since an Adeq.Prec. value greater than 4 is desirable [46]. A very low MRPD value of ±1.8% indicates an excellent agreement between predicted and actual values of FAME content. 3.1.3. Optimization of FAME content Fig. 1 shows the FAME content as a function of reaction temperature and methanol-to-oil molar ratio with a catalyst loading of 3% in the form of a response surface and contour plot based on the second-order polynomial equation. The response surface graph is suitable for optimizing the operating variables and thus for defining the optimal reaction conditions ensuring the maximum FAME content, while the contour plot is useful for understanding the interaction of the two test variables and determining their optimum levels when holding the third variable at the constant level. Fig. 1a indicates that FAME content shows a maximum for a set of reaction temperature/methanol-to-oil molar ratio combinations. FAME content increases rapidly with increasing reaction temperature from 45 °C to 55 °C and very slightly with increasing methanol-to-oil molar ratio from 9:1 to 15:1. The shape of the contours in Fig. 1b confirms a weak interaction between reaction temperature and methanol-to-oil molar ratio. The surface confined by the 99% lines in the contour plot indicates the area of maximum predicted values. It is obvious that FAME content higher than 99% can be achieved at reaction temperatures above about 56– 57 °C and any methanol-to-oil molar ratio in the range from 9:1 to 15:1. FAME content was optimized using the developed RSM model (with all actual factors) with the desired target of achieving the maximum FAME content. The maximum predicted value of the FAME content (99.8% in 4 h of reaction time) is obtained under the following reaction conditions: reaction temperature of 65 °C, methanol-to-oil molar ratio of 9:1 and catalyst loading of 3% (based on the oil weight). The achievement of the maximum FAME content at the temperature close to boiling point of methanol is attributed to the large activation energy of the reaction [41]. 3.2. Kinetics of sunflower oil methanolysis reaction 3.2.1. Analysis of sunflower methanolysis Fig. 2 illustrates variations of FAME content with the progress of sunflower oil methanolysis over the PKSB catalyst at various reaction temperatures and methanol-to-oil molar ratios at constant PKSB catalyst loadings of 3%, 5% and 7% (based on the oil weight). Both illustrations confirm that PKSB catalyst loading does not influence FAME content significantly during the reaction. The positive effect of the increase in reaction temperature in the range between 45 °C and 65 °C at the constant methanol-to-oil molar ratio of 12:1 is easily observed by comparing the curves shown in Fig. 2a, b and c. The increase of reaction temperatures accelerates the forward reaction, thus promoting FAME formation. By increasing the methanol-to-oil molar ratio at the constant reaction temperature of 45 °C, the curves slightly shift toward a shorter reaction time, thus indicating a slight increase of the FAME formation rate (Fig. 2d, a and e for molar ratios of 9:1, 12:1 and 15:1, respectively). This increase in the rate of the methanolysis reaction is more easily observed when the methanol-to-oil molar ratio increases from 9:1 to 12:1, rather than from 12:1 to 15:1, which is in accordance with the optimization results. The same trend is observed for the reactions at higher temperatures (55 °C and 65 °C).

M.D. Kostic´ et al. / Fuel 163 (2016) 304–313

309

Fig. 1. (a) Response surface, and (b) contour plot for FAME content as a function of reaction temperature and methanol-to-oil molar ratio at the catalyst loading of 3% (based on the oil weight).

Fig. 2. Variation of FAME content with time at catalyst loadings of 3% (d), 5% (N) and 7% (j) (based on the oil weight) and methanol-to-oil molar ratio of 12:1 at various reaction temperatures: (a) 45 °C, (b) 55 °C and (c) 65 °C, as well as at reaction temperature of 45 °C and various methanol-to-oil molar ratios: (d) 9:1 and (e) 15:1.

3.2.2. Kinetic modeling of methanolysis reaction The three kinetic parameters of the proposed model given by Eq. (3), namely km, K and cC0, were determined by the nonlinear least squares regression analysis using the Polymath software.

The apparent reaction rate constant, km, increases with increasing the initial methanol concentration linearly:

km ¼ k  cB0

ð8Þ

M.D. Kostic´ et al. / Fuel 163 (2016) 304–313

310

where k is the reaction rate constant and cB0 is the initial methanol concentration. Hence, the methanolysis reaction is the first order with respect to methanol. R2-values of 0.998, 0.975 and 0.962 for the linear dependences at 45, 55 and 65 °C, respectively indicate an excellent linear correlation between the apparent reaction rate constant and the initial methanol concentration. The slope of the linear dependences is the reaction rate constant, k, which depends on reaction temperature according to the Arrhenius equation: 14

k ¼ 2:51  10

  13086  exp  T

ð9Þ

The high activation energy of 108.8 kJ/mol is attributed to the three-phase heterogeneity of the reaction system resulting in a higher mass transfer limitation [41,42]. This activation energy is much higher than those for homogeneous and heterogeneous biphasic reaction systems such as the acid-catalyzed soybean oil methanolysis (27.5 kJ/mol) [47] and the esterification of oleic acid with methanol over the solid tungstophosphoric acid/SBA15 catalyst (44.6 kJ/mol) [48]. This comparison reveals that the activation energy is adversely affected by the three-phase heterogeneity of the reaction system (liquid–liquid–solid). This effect has recently been explained by the existence of enhanced mass transfer limitations in the three-phase reaction system, compared to the liquid– liquid system [41,42]. The activation energy for the sunflower oil methanolysis catalyzed by the CaO-based PKSB catalyst (108.8 kJ/mol) is found within the range (29136 kJ/mol) reported for the methanolysis of vegetable oils over various CaO-based catalysts including neat, doped, supported and mixed CaO, as can be seen in Table 3. The high activation energy means that the sunflower oil methanolysis catalyzed by the CaO-based PKSB catalyst is sensitive to temperature, which is confirmed by the high optimum reaction temperature close to the boiling point of methanol. Compared to neat CaO and CaO-based woody biomass ash, the activation energy in the present study is higher than those reported for the pseudofirst order methanolysis of soybean [41,49], palm [33], waste frying [21] (79–84 kJ/mol) and jatropha (29.5 kJ/mol) [3] oil, but it is similar to that of the nano-CaO catalyzed methanolysis of canola oil

(102–136 kJ/mol) [50]. However, the activation energy of the methanolysis of various vegetable oils over doped [16,51–53], supported [54,56] and mixed [26] CaO is much lower (16–54 kJ/mol) than that observed in the present study. The kinetic parameter K depends on both the initial methanol concentration and the reaction temperature. The dependence of K on initial methanol concentration, cB0, is linear with an R2 of 0.907, 0.972 and 0.955 for straight lines at 45, 55 and 65 °C, respectively:

K ¼ K 0  cB0

ð10Þ 0

where the constant of proportionality, K , is also a linear function of temperature (R2 = 0.967):

K 0 ¼ 7:46  104  T

ð11Þ

where T is the reaction temperature (in °C). The third kinetic parameter, cC0, does not depend on methanol concentration but decreases with increasing reaction temperature. It has already been observed that cC0 is constant at a reaction temperature if the catalyst amount is higher than 2.5% (based on the oil mass) [29]. The dependence of cC0 on temperature is an exponential function (R2 = 0.941):

cC0 ¼ 3:15  1031  T 19:31

ð12Þ

where T is reaction temperature (in °C). This parameter is related to the initial available active catalyst surface, which is, according to the applied kinetic model, related to the FAME concentration. Acting as a cosolvent, the FAMEs reduce mass transfer limitations in the three-phase system by enhancing the miscibility of sunflower oil and methanol and consequently the interfacial area. Thus, the frequency of passage of catalyst particles through the interface and thereby the available active catalyst surface is increased. In addition, intermediate products stabilize the emulsion of fine alcoholic droplets, contributing to increasing the specific liquid–liquid interfacial area. At higher reaction temperatures, the miscibility of reactants and the FAME (also intermediates) increases, thus the initial available active catalyst needed to overcome the mass transfer limitations decreases. This explains the observed negative effect of reaction temperature to cC0.

Table 3 A review on activation energy of methanolysis of vegetable oils over CaO-based catalysts. Feedstock

Catalyst

Catalyst loadinga (wt%)

Temperature (°C)

Methanol/oil (mol/mol)

Reaction order

Activation energy (kJ/mol)

Reference

Sunflower oil Palm oil Jatropha oil Soybean oil Jatropha oil, crude Soybean oil Waste frying oil Canola oil

CaO-based PKSB CaO-based woody biomass ash Lemna perpusilla Torrey ash CaOb CaOb CaO CaO CaOc

3–7 1–10 1–10 1–7 1–7 2 2 2–4

45–65 45–70 65 40–65 40–65 40–60 50–65 50–65

6:1–18:1 6:1–40:1 9:1 6:1–18:1 6:1–18:1 12:1 6.03:1 6:1–12:1

Pseudo-first Pseudo-first Pseudo-first Pseudo-first Third Pseudo-first Pseudo-first Pseudo-first

This work [33] [3] [41] [42] [49] [21] [50]

CaOd Zr/CaO KF/CaO/NiO Zn/CaO K/CaO CsF/CaO CaO/Al2O3 CaO/MgO CaO/Fly ash CaOZnO

0.1–0.5 1–6 5–20 1–10 7.5 1–10 1–6 – 2–8 2

35–65 35–75 35–65 35–65 35–65 125–200 – 45–65 60–96

15:1 3:1–18:1 3:1–18:1 12:1 6:1–12:1 6:1–12:1 – 6:1–18:1 10:1

Pseudo-first Pseudo-first Pseudo-first Pseudo-first Second, reversible Pseudo-first Second Pseudo-first Pseudo-first

108.8 83.9 29.49 82.3 133.5 81.09 79 104.83 (0–60 min) 136.48 (60–120 min) 102.49 29.8 41.2 43 54 77.92 30.7 29.31 42.56 26.5 16.6

Jatropha oil Waste cottonseed oil Waste cottonseed oil Waste cottonseed oil Soybean oil Soybean oil Rapeseed oil Crude palm oil Sunflower oil Used vegetable oil a b c d

With respect to mass of oil. Ultrasound-assisted. Nano-CaO. Higher surface area nano-CaO.

[51] [52] [53] [25] [16] [54] [55] [56] [26]

M.D. Kostic´ et al. / Fuel 163 (2016) 304–313

3.2.3. Simulation of sunflower oil methanolysis In order to verify the proposed kinetic model, the TAG conversion degree was calculated by Eq. (3) using the Polymath software and compared with the experimental data. In this computation, the values of the kinetic parameters, namely k, K and cC0, were calculated by Eqs. (8)–(12) from corresponding values of the initial methanol concentration and the reaction temperature. It was found that the MRPD between the predicted and experimental values of xA was ±25.5% (based on 117 data points), confirming the validity of the kinetic model. Deviations of the model from the experiment were observed at lower xA-values (<0.1). If these low xA-values were not considered in the comparison, then the MRPD between the predicted and experimental values of xA was only ±9.5% (based on 95 data points). Also, Fig. 3a, where variations of TAG conversion degree during the methanolysis reaction are shown, compares the predicted xA-values with the experimental data. For clarity of the illustration, only the xA-variations during the methanolysis reaction at 45 °C and 65 °C are shown. A good agreement between the model and the experimental data is obvious. Also, variations of TAG and FAME concentrations during the methanolysis under the optimum reaction conditions (temperature: 65 °C, methanol-to-oil molar ratio: 9:1, and catalyst amount: 3% based on the oil mass) calculated from the proposed kinetic model were compared with the experimental data in Fig. 3b, which confirms the agreement between the kinetic model and the experiments. The TAG and FAME concentrations were calculated from the corresponding values of the TAG conversion degrees. The important result of this kinetic study is the fact that the kinetic model originally developed for the sunflower oil methanolysis catalyzed by quicklime (basically CaO) [29] is also valid for the CaO-based PKSB methanolysis of sunflower oil. This model combines the changing- and first-order reaction rate law with respect to TAG and FAME, respectively, as well as the TAG mass transfer limitation. The changing reaction mechanism means that the ratecontrolling step involves the association of TAG with active sites present in limited but fixed amounts on the catalyst surface, while the TAG mass transfer limitation includes the autocatalytic behavior of the methanolysis reaction that increases the available active catalyst surface due to the self-enhanced specific liquid–liquid interfacial area [28]. The validity of the present kinetic model for two CaO-based catalysts (quicklime and PKSB) might indicate its generality, which should be checked in the future.

311

3.3. Catalyst reusability and calcium leaching For industrial application, not only is the catalytic activity of the PKSB catalyst important, but also its reusability. Catalyst reusability was tested under the optimum reaction conditions (reaction temperature of 65 °C, methanol-to-oil molar ratio of 9:1 and catalyst loading of 3% and reaction time of 4 h). After completion of the transesterification reaction, the PKSB catalyst was separated by centrifugation (3500 rpm, 15 min) and reused without any treatment (no washing and no recalcination). As is known, CaO can be reused without [57] or with [58] regeneration of catalytic activity. The PKSB catalyst was reused three times with little drop in catalytic activity, FAME contents of 99.0%, 98.8% and 96.2% were achieved in successive runs under optimum reaction conditions. In the fourth run, the PKSB catalyst lost its catalytic activity as the FAME content in the reaction mixture was unacceptably low (2.2%). Similarly, palm mill fly ash supported CaO has been reported to be reused three times after methanol washing and calcination (850 °C, 2 h) after each cycle, followed by a significant drop in the ester conversion in the fourth cycle [56]. Also, another ash of plant origin lost its catalytic activity in the second and third cycles of Jatropha oil methanolysis [3]. The loss of catalytic activity after a number of successive runs could be attributed to catalyst deterioration due to poisoning by glycerol [59]. Calcium glyceroxide, formed through the reaction of CaO and glycerol, could possibly be a factor reducing reusability [60]. The drawback of using CaO-based catalysts is leaching during the reaction, which is not only inversely associated with their reusability, but also contaminates the reaction products. The calcium contents in the crude FAME phases in three successive reaction runs were 722, 274 and 133 ppm. Since the calcium content was low, calcium leaching did not affect the catalyst activity. The observed calcium concentration in the final ester product was larger than those found in the FAME obtained by the methanolysis of sunflower oil (325 ppm) [61] and rapeseed oil (139 ppm) [62] catalyzed by neat CaO. The observed calcium content was higher than the EN 14214 standard limit of maximum 5 ppm, thus requiring the purification of crude biodiesel. For the purpose of comparison, three methods were employed to purify a crude biodiesel (calcium content: 659 ppm): (a) wet washing using distilled water, (b) wet washing using acidified water for neutralizing crude biodiesel followed by

Fig. 3. Comparison of (a) the predicted xA-values with the experimental data (PKSB catalyst: 3–7% based to oil mass, reaction temperature: 45 °C – open symbols and 65 °C – black symbols; methanol-to-oil molar ratio: 9:1 – circles, 12:1 – triangles and 15:1 – squares) and (b) TAG and FAME concentrations (cA – d and cC – s, respectively) calculated by the kinetic model with the experiment under the optimum reaction conditions (temperature: 65 °C, methanol-to-oil molar ratio: 9:1 and catalyst amount: 3% based on the oil mass).

M.D. Kostic´ et al. / Fuel 163 (2016) 304–313

312

Table 4 Physico-chemical properties of biodiesel obtained by methanolysis of sunflower oil over CaO-based catalysts. Property

CaO-based catalysts

EN 14214

PKSB catalyst

Neat CaO

Neat CaO

CaO nano-25%/NaX

Density (15 °C) (kg/m3) Viscosity (40 °C) (mm2/s) Acid value (mg KOH/g) Iodine value (g I2/100 g) Water (mg/kg) Ca + Mg (ppm) Cetane number FAME (%) MAG (%) DAG (%) TAG (%)

875 3.9 0.49 130 357 33 52.3 98.8 0.7 0.4 0.1

887 4.08 0.06 131 – – 46.6 99.5 – – –

– – – 123 – <14 – – 0.12 0.04 <0.05

– 4.72 0.02 – – – – 94.1 – – –

References

This work

[64]

[65]

[66]

water rinsing and (c) treatment of crude biodiesel with methanol containing anhydrous sodium carbonate at the boiling temperature for 4 h according to the procedure of Alba Rubio and coworkers [39] described in the Section 2.4. None of the applied methods resulted in the purified FAME product that satisfied the standard limit regarding calcium content (max 5 ppm). While it could be expected for the two water washing methods (calcium content of and 398 ppm, i.e. removal degree of 19% and 40% without and with neutralization, respectively), unexpectedly, the Alba-Rubio method, although most efficient (achieving removal degree of 95%; calcium content of 33 ppm), was also unsuccessful in calcium removal with respect to the prescribed limit. Both water washing methods resulted in the acid value (0.61 and 1.56 mg KOH/g without and with neutralization, respectively) above the limit (max 0.5 mg KOH/g), while the Alba-Rubio method produced the purified product with the allowed acid value (0.49 mg KOH/g). The FAME content in all three purified FAME products was higher than 97% and above the prescribed limit (96.5%), but the Alba-Rubio method produced the purest product (98.8% FAME). Therefore, the FAME obtained by the methanolysis of sunflower oil over the CaO-based PKSB catalyst and purified by the Alba-Rubio method requires further purification to lower the calcium level below the specified limit. Special attention is being paid to extracting the leached calcium from the crude biodiesel produced with calcium oxide as a catalyst [63]. The physico-chemical properties of the purified biodiesel, along with the data reported in the previous studies, are summarized in Table 4. For comparison, the biodiesel standard EN 14214 specifications are also given. The properties are mainly within the biodiesel quality standard EN14214. Key biodiesel physicochemical properties including ester and water contents, acid value, kinematic viscosity, density and cetane number are within the limits specified by the EN 14214. The calcium content, iodine value and DAG content are above the specified limits. Since the calcium and DAG contents in the biodiesel product are higher than the standard limit, the overall process should include an adequate purification stage. The iodine value of the biodiesel is higher than the standard limit, but there is evidence that a biodiesel with an iodine value of 130 g I2/100 g of oil would be oxidatively stable [64]. 4. Conclusion The potential of the low-cost basic catalyst derived from palm kernel shell gasification residues is demonstrated for the in transesterification reactions by coupling full factorial design with modeling, optimization, and kinetic studies. The statistical analysis predicts that the effect of reaction temperature and methanol-tooil molar ratio on the FAME conversion is significant, while the

860–900 3.5–5.0 0.50 max 120 max 500 max 5 51 min 96.5 min 0.8 max 0.2 max 0.2 max

effect of catalyst loading is statistically negligible. Under the optimum process conditions, almost complete TAG conversion is achieved within 4 h. The same kinetic model already used for sunflower methanolysis over several CaO-based catalysts is shown to be valid for the PKSB. However, since the CaO-based PKSB catalyst can be reused in only three consecutive runs without any treatment, the possibility of regeneration of the spent catalyst by appropriate washing and recalcination must be tested. One disadvantage of the catalyst is that the calcium leaches into the biodiesel product, which consequently requires the inclusion of an additional purification stage in the overall process. Overall, the utilization of the PKSB in biodiesel production not only reduces the catalyst cost but also offers a cost-effective way of recycling this waste, thus reducing the environmental impact of its disposal. The fact that PKSB is a waste, compensates the limited reusability of the PKSB catalyst. Acknowledgments This work has been jointly funded by the Ministry of Education, Science and Technological Development of the Republic of Serbia (Project III 45001), and the Hong Kong University of Science and Technology, School of Engineering (SENG-HKRGC). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.fuel.2015.09.042. References [1] Veljkovic´ VB, Stamenkovic´ OS, Tasic´ MB. The wastewater treatment in the biodiesel production with alkali-catalyzed transesterification. Renew Sust Energy Rev 2014;32:40–60. [2] Sharma M, Khan AA, Puri SK, Tuli DK. Wood ash as a potential heterogeneous catalyst for biodiesel synthesis. Biomass Bioenergy 2012;41:94–106. [3] Chouhan APS, Sarma AK. Biodiesel production from Jatropha curcas L. oil using Lemna perpusilla Torrey ash as heterogeneous catalyst. Biomass Bioenergy 2013;55:386–9. [4] Vadery V, Narayanan BN, Ramakrishnan RM, Cherikkallinmel SK, Sugunan S, Narayanan DP, et al. Room temperature production of jatropha biodiesel over coconut husk ash. Energy 2014;70:588–94. [5] Bazargan A, Kostic´ MD, Stamenkovic´ OS, Veljkovic´ VB, McKay G. Palm kernel shell gasification residues as precursors for a calcium oxide-based catalyst for biodiesel production. Fuel 2015;150:519–25. [6] Kouzu M, Hidaka J, Komichi Y, Nakano H, Yamamoto M. A process to transesterify vegetable oil with methanol in the presence of quick lime bit functioning as solid base catalyst. Fuel 2009;88:1983–90. [7] Miladinovic´ MR, Stamenkovic´ OS, Veljkovic´ VB, Skala DU. Continuous sunflower oil methanolysis over quicklime in a packed-bed tubular reactor. Fuel 2015;154:301–7. [8] Boey PL, Ganesan S, Maniam GP, Khairuddean M. Catalysis derived from waste sources in the production of biodiesel using waste cooking oil. Catal Today 2012;190:117–21.

M.D. Kostic´ et al. / Fuel 163 (2016) 304–313 [9] Chakraborty R, Bepari S, Benerjee A. Transesterification of soybean oil catalyzed by fly ash and egg shell derived solid catalysts. Chem Eng J 2010;165:798–805. [10] Dias JM, Alvim-Ferraz MCM, Almeida MF, Diaz JDM, Polo MS, Utrilla JR. Selection of heterogeneous catalysts for biodiesel production from animal fat. Fuel 2012;94:418–25. [11] Liao CC, Chung TW. Optimization of process conditions using response surface methodology for the microwave-assisted transesterification of Jatropha oil with KOH impregnated CaO as catalyst. Chem Eng Res Des 2013;91:2457–64. [12] Mahdavi V, Monajemi A. Optimization of operational conditions for biodiesel production from cottonseed oil on CaO–MgO/Al2O3 solid base catalysts. J Taiwan Inst Chem Eng 2014;45:2286–92. [13] Nakatani N, Takamori H, Takeda K, Sakugawa H. Transesterification of soybean oil using combusted oyster shell waste as a catalyst. Bioresour Technol 2009;100:1510–3. [14] Zabeti M, Daud WMAW, Aroua MK. Biodiesel production using aluminasupported calcium oxide: an optimization study. Fuel Process Technol 2010;91:243–8. [15] Kaur M, Ali A. Ethanolysis of waste cottonseed oil over lithium impregnated calcium oxide: kinetics and reusability studies. Renew Energy 2014;63:272–9. [16] Liu C-C, Lu W-C, Liu T-J. Transesterification of soybean oil using CsF/CaO catalysts. Energy Fuel 2012;26:5400–7. [17] Wang B, Li S, Tian S, Feng R, Meng Y. A new solid base catalyst for the transesterification of rapeseed oil to biodiesel with methanol. Fuel 2013;104: 698–703. [18] Wu H, Zhang J, Wei Q, Zheng J, Zhang J. Transesterification of soybean oil to biodiesel using zeolite supported CaO as strong base catalysts. Fuel Process Technol 2013;109:13–8. [19] Patil P, Gude VG, Pinappu S, Deng S. Transesterification kinetics of Camelina sativa oil on metal oxide catalysts under conventional and microwave heating conditions. Chem Eng J 2011;168:1296–300. [20] Singh AK, Fernando SD. Reaction kinetics of soybean oil transesterification using heterogeneous metal oxide catalysts. Chem Eng Technol 2007;12: 1716–20. [21] Birla A, Singh B, Upadhyay SN, Sharma YC. Kinetics studies of synthesis of biodiesel from waste frying oil using a heterogeneous catalyst derived from snail shell. Bioresour Technol 2012;106:95–100. [22] Deshmane VG, Adewuyi YG. Synthesis and kinetics of biodiesel formation via calcium methoxide base catalyzed transesterification reaction in the absence and presence of ultrasound. Fuel 2013;107:474–82. [23] Ilgen O. Reaction kinetics of dolomite catalyzed transesterification of canola oil and methanol. Fuel Process Technol 2012;95:62–6. [24] Kouzu M, Kasuno T, Tajika M, Sugimoto Y, Yamanaka S, Hidaka J. Calcium oxide as a solid base catalyst for transesterification of soybean oil and its application to biodiesel production. Fuel 2008;87:2798–806. [25] Kumar D, Ali A. Nanocrystalline K-CaO for the transesterification of a variety of feedstocks: structure, kinetics and catalytic properties. Biomass Bioenergy 2012;46:459–68. [26] Lukic´ I, Kesic´ Zˇ, Maksimovic´ S, Zdujic´ M, Liu H, Krstic´ J, et al. Kinetics of sunflower and used vegetable oil methanolysis catalyzed by CaOZnO. Fuel 2013;113:367–78. [27] Lukic´ I, Kesic´ Zˇ, Maksimovic´ S, Zdujic´ M, Krstic´ J, Skala D. Kinetics of heterogeneous methanolysis of sunflower oil with CaOZnO catalyst: influence of different hydrodynamic conditions. Chem Ind Chem Eng Q 2014;20:425–39. [28] Veljkovic´ VB, Stamenkovic´ OS, Todorovic´ ZB, Lazic´ ML, Skala DU. Kinetics of sunflower oil methanolysis catalyzed by calcium oxide. Fuel 2009;88: 1554–62. [29] Miladinovic´ MR, Krstic´ JB, Tasic´ MB, Stamenkovic´ OS, Veljkovic´ VB. A kinetic study of quicklime-catalyzed sunfloweroil methanolysis. Chem Eng Res Des 2014;92:1740–52. [30] Tasic´ MB, Miladinovic´ MR, Stamenkovic´ OS, Veljkovic´ VB, Skala DU. Kinetic modeling of sunflower oil methanolysis catalyzed by calcium-based catalysts. Chem Eng Technol 2015;38:1550–6. [31] Miladinovic´ MR, Tasic´ MB, Stamenkovic´ OS, Veljkovic´ VB, Skala DU. Further study on kinetic modeling of sunflower oil methanolysis catalyzed by calciumbased catalysts. Chem Ind Chem Eng Q 2015. http://dx.doi.org/10.2298/ CICEQ150618027M. [32] Lee SL, Wong YC, Tan YP, Yew SY. Transesterification of palm oil to biodiesel by using waste obtuse horn shell-derived CaO catalyst. Energy Convers Manage 2015;93:282–8. [33] Maneerung T, Kawi S, Wang C-H. Biomass gasification bottom ash as a source of CaO catalyst for biodiesel production via transesterification of palm oil. Energy Convers Manage 2015;92:234–43. [34] Sánchez M, Marchetti JM, El Boulifi N, Aracil J, Martínez M. Kinetics of Jojoba oil methanolysis using a waste from fish industry as catalyst. Chem Eng J 2015;262:640–7. [35] Rezaei R, Mohadesi M, Moradi GR. Optimization of biodiesel production using waste mussel shell catalyst. Fuel 2013;109:534–41. [36] Hu S, Wang Y, Han H. Utilization of waste freshwater mussel shell as an economic catalyst for biodiesel production. Biomass Bioenergy 2011;35:3627–35. [37] Moradi G, Mohammadi F. Utilization of waste coral for biodiesel production via transesterification of soybean oil. Int J Environ Sci Technol 2014;11: 805–12. [38] Correia LM, Saboya RMA, de Sousa Campelo N, Cecilia JA, Rodríguez-Castellón E, Cavalcante Jr CL, et al. Characterization of calcium oxide catalysts from

[39]

[40]

[41]

[42]

[43]

[44] [45]

[46]

[47]

[48]

[49]

[50] [51]

[52]

[53]

[54]

[55]

[56]

[57] [58]

[59]

[60]

[61] [62]

[63]

[64]

[65]

[66]

313

natural sources and their application in the transesterification of sunflower oil. Bioresour Technol 2014;151:207–13. Alba-Rubio AC, Alonso Castillo ML, Albuquerque MCG, Mariscal R, Cavalcante Jr CL, Granados ML. A new and efficient procedure for removing calcium soaps in biodiesel obtained using CaO as a heterogeneous catalyst. Fuel 2012;95: 464–70. Avramovic´ JM, Stamenkovic´ OS, Todorovic´ ZB, Lazic´ ML, Veljkovic´ VB. Empirical modeling the ultrasound-assisted base-catalyzed sunflower oil methanolysis kinetics. Chem Ind Chem Eng Q 2012;18:115–27. Choudhury HA, Chakma S, Moholkar VS. Mechanistic insight into sonochemical biodiesel synthesis using heterogeneous base catalyst. Ultrason Sonochem 2014;21:169–81. Choudhury HA, Goswami PP, Malani RS, Moholkar VS. Ultrasonic biodiesel synthesis from crude Jatropha curcas oil with heterogeneous base catalyst: mechanistic insight and statistical optimization. Ultrason Sonochem 2014;21: 1050–64. Yan S, Lu H, Liang B. Supported CaO catalysts used in the transesterification of rapeseed oil for the purpose of biodiesel production. Energy Fuel 2008;22: 646–51. Yang Z, Xie W. Soybean oil transesterification over zinc oxide modified with alkali earth metals. Fuel Process Technol 2007;88:631–8. Myers RH, Montgomery DC. Response surface methodology: process and product optimization using designed experiments. 2nd ed. New York: John Wiley and Sons; 2002. Anderson MJ, Whitcomb PJ. RSM simplified: optimizing processes using response surface methods for design of experiments. New York: Productivity Press; 2005. Parkar PA, Choudhary HA, Moholkar VS. Mechanistic and kinetic investigations in ultrasound assisted acid catalyzed biodiesel synthesis. Chem Eng J 2012;187:248–60. Brahmkhatri V, Patel A. 12-Tungstophosphoric acid anchored to SBA-15: an efficient, environmentally benign reusable catalyst for biodiesel production by esterification of free fatty acids. Appl Catal A 2011;403:161–72. Liu X, Piao X, Wang Y, Zhu S. Model study on transesterification of soybean oil to biodiesel with methanol using solid base catalyst. J Phys Chem A 2014;110: 3750–5. Zhao L, Qiu Z, Stagg-Williams SM. Transesterification of canola oil catalyzed by nanopowder calcium oxide. Fuel Process Technol 2013;114:154–62. Kaur N, Ali A. Kinetics and reusability of Zr/CaO as heterogeneous catalyst for the ethanolysis and methanolysis of Jatropha crucas oil. Fuel Process Technol 2014;2014(119):173–84. Kaur M, Ali A. Potassium fluoride impregnated CaO/NiO: an efficient heterogeneous catalyst for transesterification of waste cottonseed oil. Eur J Lipid Sci Technol 2014;116:80–8. Kumar D, Ali A. Transesterification of low-quality triglycerides over a Zn/CaO heterogeneous catalyst: kinetics and reusability studies. Energy Fuel 2013;27: 3758–68. Pasupulety N, Gunda K, Liu Y, Rempel GL, Ng FTT. Production of biodiesel from soybean oil on CaO/Al2O3 solid base catalysts. Appl Catal A – Gen 2013;452: 189–202. Yao J, Li W, Yao C, Guo D, Yang H. Kinetics of transesterification of rapeseed oil by methanol catalyzed by solid base catalyst calcium oxide/magnesium oxide. Spec Petrochem 2008;25:39–42. Ho WWS, Ng HK, Gan S, Tan SH. Evaluation of palm oil mill fly ash supported calcium oxide as a heterogeneous base catalyst in biodiesel synthesis from crude palm oil. Energy Convers Manage 2014;88:1167–78. Wei Z, Xu C, Li B. Application of waste eggshell as low-cost solid catalyst for biodiesel production. Bioresour Technol 2009;100:2883–5. Boey PL, Maniam GP, Hamid SA. Biodiesel production via transesterification of palm olein using waste mud crab (Scylla serrata) shell as a heterogeneous catalyst. Bioresour Technol 2009;100:6362–8. Liu X, Piao X, Wang Y, Zhu S. Calcium ethoxide as a solid base catalyst for the transesterification of soybean oil to biodiesel. Energy Fuels 2008;22: 1313–7. Kouzu M, Kasuno T, Tajika M, Yamanaka S, Hidaka J. Active phase of calcium oxide used as solid base catalyst for transesterification of soybean oil with refluxing methanol. Appl Catal A 2008;334:357–65. Verziu M, Coman SM, Richards R, Parvulescu VI. Transesterification of vegetable oils over CaO catalysts. Catal Today 2011;167:64–70. Kouzu M, Yamanaka S, Hidaka J, Tsunomori M. Heterogeneous catalysis of calcium oxide used for transesterification of soybean oil with refluxing methanol. Appl Catal A – Gen 2009;355:94–9. Veljkovic´ VB, Bankovic´-Ilic´ IB, Stamenkovic´ OS. Purification of crude biodiesel obtained by heterogeneously-catalyzed transesterification. Renew Sust Energy Rev 2015;49:500–16. Predojevic´ Z, Škrbic´ B, Šadurišic´-Mladenovic´ N. Transesterification of linoleic and oleic sunflower oils to biodiesel using CaO as a solid base catalyst. J Serb Chem Soc 2012;77:815–32. Son SM, Kusakabe K, Guan G. Biodiesel synthesis and properties from sunflower and waste cooking oils using CaO catalyst under reflux conditions. J Appl Sci 2010;10:3191–8. Luz Martínez S, Romero R, López JC, Romero A, Sánchez Mendieta V, Natividad R. Preparation and characterization of CaO nanoparticles/NaX zeolite catalysts for the transesterification of sunflower oil. Ind Eng Chem Res 2011;50:2665–70.