Scholarly article on topic 'Influence of operating variables on the transesterification of waste cooking oil to biodiesel over sodium silicate catalyst: A statistical approach'

Influence of operating variables on the transesterification of waste cooking oil to biodiesel over sodium silicate catalyst: A statistical approach Academic research paper on "Chemical sciences"

CC BY-NC-ND
0
0
Share paper
OECD Field of science
Keywords
{Biodiesel / "Waste cooking oil" / Transesterification / "Heterogeneous catalysis"}

Abstract of research paper on Chemical sciences, author of scientific article — M.O. Daramola, K. Mtshali, L. Senokoane, O.M. Fayemiwo

Abstract This study examined the use of surface response methodology to investigate the influence of operating variables on the transesterification of waste cooking oil (WCO) to biodiesel over sodium silicate catalysts. The individual and interactive effects of three variables namely, reaction time, reaction temperature and amount of catalyst was evaluated using full 23 (+1) factorial design. The conversion of WCO to biodiesel was achieved through the transesterification reaction over the catalyst at a methanol-to-oil molar ratio of 6:1 in a batch reactor. Physicochemical properties of the sodium silicate catalyst were obtained using Fourier transform infrared spectroscopy (FT-IR) for surface chemistry, thermo-gravimetric analysis (TGA) for thermal stability, N2 physisorption test for Brunauer–Emmett–Teller analysis and scanning electron microscopy (SEM) for morphology. The reaction temperature, reaction time and weight of the catalyst (expressed as a percentage of the amount of WCO) were varied to understand their effect on the yield of biodiesel via response surface methodology (RSM) approach. The BET analysis showed a surface area of 0.386m2/g for the catalyst. Results from the transesterification reaction reveal that change in catalyst weight percentage had no considerable effect on the biodiesel yield and that there was no mutual interaction between the reaction time and catalyst weight percentage. The results also conveyed that the reaction temperature and reaction time were limiting conditions and a slight variation herein altered the biodiesel yield. The transesterification of WCO produced 57.92% maximum FAME yield at the optimum methanol to oil molar ratio of 6:1, catalyst weight of 2.5%, reaction time of 240min and a reaction temperature of 64°C. The variance ratio, VR < F value obtained from the cross-validation experiments indicate perfect agreement of the model output with experimental results and also testifies to the validity and suitability of the model to predict the biodiesel yields.

Academic research paper on topic "Influence of operating variables on the transesterification of waste cooking oil to biodiesel over sodium silicate catalyst: A statistical approach"

lIlP

Available online atwww.sciencedirect.com

ScienceDirect

Journal of Taibah University for Science xxx (2015) xxx-xxx

of Taibah University

for Science ournal

www.elsevier.com/locate/jtusci

Influence of operating variables on the transesterification of waste cooking oil to biodiesel over sodium silicate catalyst: A statistical approach

M.O. Daramola *, K. Mtshali, L. Senokoane, O.M. Fayemiwo

School of Chemical and Metallurgical Engineering, Faculty of Engineering and the Built Environment, University of the Witwatersrand,

Wits 2050, Johannesburg, South Africa

Received 29 April 2015; received in revised form 2 July 2015; accepted 28 July 2015

Abstract

This study examined the use of surface response methodology to investigate the influence of operating variables on the transesterification of waste cooking oil (WCO) to biodiesel over sodium silicate catalysts. The individual and interactive effects of three variables namely, reaction time, reaction temperature and amount of catalyst was evaluated using full 23 (+1) factorial design. The conversion of WCO to biodiesel was achieved through the transesterification reaction over the catalyst at a methanol-to-oil molar ratio of 6:1 in a batch reactor. Physicochemical properties of the sodium silicate catalyst were obtained using Fourier transform infrared spectroscopy (FT-IR) for surface chemistry, thermo-gravimetric analysis (TGA) for thermal stability, N2 physisorption test for Brunauer-Emmett-Teller analysis and scanning electron microscopy (SEM) for morphology. The reaction temperature, reaction time and weight of the catalyst (expressed as a percentage of the amount of WCO) were varied to understand their effect on the yield of biodiesel via response surface methodology (RSM) approach. The BET analysis showed a surface area of 0.386 m2/g for the catalyst. Results from the transesterification reaction reveal that change in catalyst weight percentage had no considerable effect on the biodiesel yield and that there was no mutual interaction between the reaction time and catalyst weight percentage. The results also conveyed that the reaction temperature and reaction time were limiting conditions and a slight variation herein altered the biodiesel yield. The transesterification of WCO produced 57.92% maximum FAME yield at the optimum methanol to oil molar ratio of 6:1, catalyst weight of 2.5%, reaction time of 240 min and a reaction temperature of 640C. The variance ratio, VR < Fvalue obtained from the cross-validation experiments indicate perfect agreement of the model output with experimental results and also testifies to the validity and suitability of the model to predict the biodiesel yields.

©2015 The Authors. Production and hosting by Elsevier B.V. on behalf of Taibah University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Biodiesel; Waste cooking oil; Transesterification; Heterogeneous catalysis

* Corresponding author. Tel.: +27 11 717 7536; fax: +27 11 717 7536. E-mail address: Michael.daramola@wits.ac.za (M.O. Daramola). Peer review under responsibility of Taibah University.

http://dx.doi.Org/10.1016/j.jtusci.2015.07.008

1658-3655 © 2015 The Authors. Production and hosting by Elsevier B.V. on behalf of Taibah University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

2 M.O. Daramola et al. /Journal ofTaibah University for Science xxx (2015) xxx-xxx

1. Introduction

Combustion of fossil fuels has an immense detrimental effect on the environment. The release of pollutant gases such as NOx, SOx, and CO is inevitable during this process. In addition, the price instability of fossil fuels poses a serious threat to countries with limited resources. Taking into account the limited amount of energy resources, their increasing prices and environmental issues associated with the use of conventional fuels for energy, other means of producing energy sus-tainably have become the forefront of research. Amongst the studies for other means of providing sustainable energy is biodiesel, which is a branch of biofuels and has become an essential alternative for liquid fuels.

Previous studies have demonstrated the use of typical edible plant oils, such as soybean, rapeseed oil and palm oil for the production of biodiesel [1]. These raw materials are not entirely suitable, more especially in developing countries, due to the limited supply and high cost associated with their application, as well as competition with the food chain [2]. Therefore, low cost, non-edible oils such as jatropha oil, animal fat and waste cooking oil have been suggested and tested as alternatives [3-5]. However, the main disadvantage of these non-edible types of feedstock is the high content of free fatty acid (FFA) within the oils, which poses problems in the production process. Consequently, Biodiesel from high FFA content feedstock is conventionally produced by a two-stage process: esterification, followed by transesterification [6,7]. The esterification step serves to reduce the amount of FFAs present in the oil in order to allow for the transesterification reaction to commence [7,8]. The implication of this additional process unit is inevitably the additional costs associated with the biodiesel production.

In recent times, the use of heterogeneous catalysts has been proven to be very effective in converting high FFA feedstock directly to biodiesel, thereby by-passing the esterification stage [3,9,10]. The most commonly used heterogeneous catalysts for the production of biodiesel are ion-exchange resins, inorganic-oxide solid acids and supported noble-metal oxides. However, a dramatic decrease in the catalytic activity of these catalysts has been observed due to their absorption of water during biodiesel production [9]. Besides the sharp reduction in the catalytic activity, the catalysts can form a slurry with the products by absorbing water and carbon dioxide, thereby increases the viscosity of the product mixture, making product separation very difficult [11]. Guo et al. [11] reported that the use of sodium silicate as

solid catalyst suppresses the formation of soap because of the decreased water content (less than 4%). Guo et al. [3,12] demonstrated the excellent performance of calcined sodium silicate catalyst for the transesteri-fication of soy bean oil to biodisel. Sodium silicate has a high catalytic activity after calcination and is immiscible with triglycerides and alcohol [3]. During trans-esterification, hydrolysis reaction with sodium silicate and water resulted in the formation of NaOH and Si-O-H [3]. Furthermore, a yield of about 97% has been reported at catalyst amount of 7 wt.% and a methanol/oil ratio of 6:1 for the transesteriftaion of soy bean oil to biodiesel using sodium silicate [12].

In spite of active research efforts in the development and use of heterogeneous catalysts for biodiesel production, only a few reports have been documented in literature related to the investigation of the influence of operating variables on the trans-esterification of waste cooking oil/soy bean oil to biodiesel over sodium silicate. Besides, most of these reports adopted a traditional approach whereby one variable is investigated at a time

[12]. This approach overlooks the interactive effect of different variables on the results. Understanding these effects requires the use of an alternative approach. As a follow-up on our recent studies on the transesterification of WCO to biodiesel over calcined sodium silicate

[13], in this article the use of reseponse surface methodology (RSM) approach to investigate effect of operating variables is presented. The variables considered were reaction time, amount of catalyst and reaction temperature while the alcohol-to-oil ratio was fixed at 6:1 following report from [12].

2. Materials and methods

2.1. Determination of the free fatty acid (FFA) content of the oil

The waste cooking oil was obtained from a food vendor at the University of the Witwatersrand, and its FFA content in WCO was determined to confirm the need for a heterogeneous catalyst such as sodium silicate. It was also required to prove the tolerance of sodium silicate to a high content of free fatty acid in biodiesel production. The free fatty acid (FFA) content of the WCO was evaluated according to the procedure described elsewhere [13]. A 1.0 ml of the WCO diluted with 10 ml of 99% isopropyl alcohol was titrated against 0.025 M NaOH solution dropwisely using phenolphthalein solution (0.05 g of phenolphthalein to 50 ml of 95% pure ethanol and diluted to a 100 ml using distilled water)

M.O. Daramola et al. /Journal ofTaibah University for Science xxx (2015) xxx-xxx

as the pH indicator. The percentage FFA was calculated according to Eq. (1):

%FFA = (V -b) xNx--(1)

where %FFA is the percentage free fatty acid (FFA) of the WCO; V, the titrant value (ml); b, the volume of the blank (ml); N, the concentration of the titration solution in mg/l, and w, the weight of the 1 ml sample of the WCO.

2.2. Catalyst characterization

The catayst, sodium metasilicate, was purchased from Sigma-Aldrich (Pty) South Africa. Physicochemical properties such as morphology, surface chemistry and thermal stability of the catalyst were determined using scanning electron microscopy (SEM), Fourier transform infrared spectroscopy (FTIR) thermogravimetric analysis (TGA), respectively. Nitrogen physisorption at 77 K under isothermal condition was conducted on the catalyst to determine BET surface area, pore volume and the pore size of the catalyst. During the physisoprtion experiment, 0.12 g of the catalyst was used. The sample was degassed before the N2 physisorption. The FT-IR analysis was conducted with a wave number range from 400 to 4000 cm-1. The SEM images were taken on a Carl Zeiss, operating at an accelerated voltage of 5.00 kV. The TGA analysis was conducted on the catalyst sample to observe its thermal stability and also to determine the temperature at which calcination should be carried out. About 0.89 g of the sample was subjected to TGA analysis under the flow of helium gas and the data was collected every 2 min.

2.3. Design of experiment and transesterifiction reaction

Prior to trans-esterification, WCO and sodium silicate were pre-treated. The WCO was preheated at a temperature of 120 °C to ensure that any trace of moisture in the raw material was removed before the transesteri-fication reaction. The sodium silicate was calcined at 200 °C for 2h to remove moisture and any organic contaminants from the catalyst. The equipment used in the trans-esterification process was a Liebig condenser that was connected to a running tap which supplied cool water to prevent the volatilization of volatile compounds from the reaction mixture as (see [13] for detail). The percentage weight of the catalyst used in the transesterification was calculated based on the weight of the WCO.

Table 1

Design of experiment using a 23(+1) full-factorial design showing coded and real variables.

Expt. run Coded values Actual values

T (°C) t (min) W T (°C) t (min) W (wt.%)

1 2 -1 -1 -1 30 30 0.5

+ 1 -1 -1 64 240 0.5

3 -1 +1 -1 30 30 0.5

4 +1 +1 -1 64 240 0.5 2.5

5 -1 -1 +1 30 30 2.5

6 +1 -1 +1 64 240 2.5

7 -1 +1 +1 30 30 2.5

8 +1 +1 +1 64 240

9 0 0 0 50 135 1.5

T: Reaction temperature (°C), t: reaction time (min); W: catalyst weight (wt.%); +1: upper limit; 0: middle point; -1: lower limit.

A two-level full factorial design (23 (+1) full-factorial) was designed according to the data provided in Table 1. The three factors that were investigated within the identified lower and higher levels were the reaction temperature (T), reaction time (t) and catalyst weight (W). For a two-level full factorial design, there are 2k different combinations of the levels. The two-three factorial method was used, implying that three factors (k) were considered and eight (8) experimental runs plus one at the centre were performed, making a total of 9 runs. A general second order regression model presented in Eq. (2) was adopted and the regression coefficients of the regression equation were estimated using the Least Square (LS) parameter estimation method (using Eq. (3)) implemented in the matlab environment. Furthermore, the model was validated using cross-validation technique, a technique that is useful to estimate how accurately a predictive model will perform in practice [14,15]. During the cross-validation an analysis is performed on a training set of data (data used to obtain the regression model) and validating the analysis on testing set of data obtained from repeated experiments using independent experimental runs as explained elsewhere [16]. In addition, one-way analysis of variance (ANOVA) was also computed to evaluate the statistical significance and validity of the model.

Y = Y0 + a\T + a2t + a3W + a4 T2 + a5i2 + a6W2 + a7Tt + a8TW + a9tW + a10TWt (2)

a = [ xt x ] 1xty (3)

where a is a (u x 1) vector of the regression coefficients; X, a (N x u) matrix of the coded levels of input variables;

M.O. Daramola et al. /Journal ofTaibah University for Science xxx (2015) xxx-xxx

XT, the transpose of X; and Y, a (N x 1) vector of the reversibility determined experimentally according to the experimental design, N.

2.4. Product (biodiesel) analysis

In the recent times, Fourier transform infrared (FT-IR) spectroscopy is employed as a modern analytical technique for the detection of biodiesel due to its rapid detection method [17]. Therefore, qualitative analysis of the biodiesel produced was carried out using FT-IR and conducted using a commercial biodiesel standard purchased from Sigma-Aldrich as a reference. Quantitatively, the biodiesel yield was calculated using Eq. (4):

Biodiesel Yield(%)

Actual Mass of Biodiesel Produced Theoretical Mass of Biodiesel Produced

3. Results and discussion

3.1. FFA content

The free fatty acid (FFA) content in the WCO sample was 3.28%. According to previous studies, the base-catalyzed transesterification requires oil samples with FFA content less than 1% to avoid the occurrence of hydrolysis and saponification reactions [18,19]. The production of biodiesel was successful even though the FFA content exceeded the recommended 1% because of the use of sodium silicate as a catalyst for transes-terification. The tolerance of sodium silicate to water is due to its polyhedron network [20] and porous structure [11], which allows for sequential hydration to occur in three steps when there are high amounts of water. When this happens, the Si-O-Si bridges hydrolyze and causes H4SiO2 monomers to be released thus producing OH-and avoiding soap formation [11].

The transesterification reaction proceeded successfully regardless of the high FFA, due to the use of the calcined sodium silicate, thereby eliminating the need for esterification reaction. The optimum biodiesel yield obtained in the study was 57.92%. This value is lower than the value reported by Guo et al. [11] for the transesterificaton of soy oil to biodiesel over calcined sodium silicate. It is noteworthy to mention that the feedstock used in this study is different from the one used in Ref. [11]. Moreoever, the relatively high FFA content of 3.28% in the feedstock

Wavenumber (cm" ) Fig. 1. FT-IR analysis of the commercial sodium silicate catalyst.

Fig. 2. SEM image of the catalyst.

used in this study could be another reason for the low yield.

3.2. Catalyst characterization

According to literature, the stretches of the Si-O-Na and Si-O-Si are expected to be located at wavelengths of ~1000 cm-1 [21] and ~1381cm-1 [22], respectively. In Fig. 1, the stretches located at wavelength ~1005cm-1,941 cm-1 and ~1400cm-1 correspond to Si-O-Na and Si-O-Si stretches [21,22], showcasing the polyhedron structure of sodium silicate. According to Hindryawati et al. [22], the Na-O bond structure vibration is characterised by absorption peaks at wavelengths that range from 486 to 619 cm-1. The FTIR spectra in Fig. 2 show an absorption band at a wavelength of ~614cm-1, confirming the sodium silicate structure according to the literature. The peaks located at ~794 cm-1 and 1101 cm-1 could be attributed to the O-Si-O stretches [22]. The broad band at ~3500cm-1 could be attributed to O-H bend and stretch associated to water molecules [22]. This observation then shows that the sodium silicate catalyst contains a bit of water which could be explained by the fact that it is hygroscopic.

M.O. Daramola et al. /Journal ofTaibah University for Science xxx (2015) xxx-xxx

Table 2

BET analysis results of commercial sodium silicate.

Parameter Value

Surface area (m2/g) 0.39

Pore volume (cm3/g) 6.24 x 10-3

Pore size (nm) 59.18

Lastly, the band located at a range of 1960-1723 cm-1 could be attributed to organics.

The SEM image of sodium silicate catalyst, which represents particles in the range of 1 ^m, is depicted in Fig. 2. According to Guo et al. [11], structures which are loosely attached and contain spaces of 1-5 ^m between agglomerates, are favorable to entry of triglyceride and methanol. This then allows for the basic sites of the surface of the catalyst to be used for transesterification. Since Fig. 2 shows that the particles observed are in range, it suggests that the commercial sodium silicate catalyst will exhibit an affinity to the entry of triglyceride and alcohol for the progression of transesterification.

The BET surface area, pore volume and the pore size of the catalyst are presented in Table 2. According to Guo et al. [11], a surface area of 5.91 m2/g of calcined sodium silicate was reported for their experiment. The results of the experiment using 3.0 wt.% of catalyst, a methanol:oil ratio of 7.5:1, a reaction time of 60 min and a temperature of 60 °C displayed a biodiesel yield of almost 100% from soy bean oil. The surface area of sodium silicate for the experiment run that resulted in one of the highest yield (Experimental runs 2) of 57.20% in this study was found to be 0.39 m2/g. The operating conditions for run 2 were a catalyst weight of 0.5%, a reaction time of 240 min and a temperature of 64 °C. A BET analysis was performed using N2 gas. The BET analysis showed that the surface area of the sodium silicate was 0.39 m2/g. A similar study reported by Guo et al. [11] using soy oil as the feedstock, used a surface area of 5.91 m2/g of sodium silicate catalyst. The study conducted aimed at producing biodiesel from soybean under the following conditions: reaction temperature of 60 °C, sodium silicate of 2.5 wt.%, a molar ratio of methanol/oil of 6:1 and reaction time of 60 min. The biodiesel yield obtained from the conditions stated was 80%. When observing the experimental design in this current study for runs 6 and 8, it is noted that the operating conditions (reaction temperature, methanol/oil molar ratio and catalyst weight a) are similar to that of Guo et al. [11]. The biodiesel yield obtained for experimental runs 6 and 8 were 54.72% and 57.92% respectively. One would expect a similar trend in biodiesel yield for the current study and that of Guo et al.

Fig. 3. Qualitative analysis of biodiesel using FTIR. Reference biodiesel (top); biodiesel from WCO (bottom).

[11], but the discrepancy could be accounted for by the marginal difference in the surface area of sodium silicate. The larger surface area of 5.91 m2/g is an indication that the sodium silicate will display a higher catalytic activity which then explains the yield of 80% reported by Guo et al. [11] when compared to this study.

3.3. Qualitative analysis of biodiesel

A quantitative analysis of the biodiesel produced was done using FT-IR spectroscopy. According to O'Donnell et al. [17], FT-IR has been employed as a modern analytical technique for the detection of biodiesel due to its rapid detection method. The analysis was conducted using a commercial biodiesel standard and products from experimental runs. The analyses are shown in Fig. 3. It is observed that the product displayed a spectrum that is more comparable to the standard biodiesel. The most characteristic peak on a biodiesel IR spectrum is one at 1200 cm-1 which is related to O-CH3 vibrations [23]. When comparing Fig. 3(top) and Fig. 3(bottom), it is evident that the peak characterised by O-CH3 vibrations is prominent in both spectra. The peak gives an indication of the attachment of the alkyl group of the alcohol to the fatty acid group in the triglyceride, thus forming an ester. From Fig. 3(top) and Fig. 3(bottom), there are bands appearing in the range of 1170-1200 cm-1, these

6 M.O. Daramola et al. /Journal ofTaibah

Table 3

The experimental and model biodiesel yield from WCO with the errors.

Experimental run Response (Y) Error % Error

Experimental Predicted

1 2 41.79 57.20 41.86 53.20 -0.08 -0.19

44.23 4.01 2.37 7.00

3 41.86 53.20 -6.29 5.35

4 46.91 42.29 -13.41

5 54.72 43.84 -1.56 -3.68

6 55.18 -0.46 -0.85

7 43.11 57.92 43.84 -0.73 2.74 -1.70

8 55.18 4.73

9 53.40 53.39 0.01 0.01

are ester peaks with a (C-O) vibration [24]. The presence of the (CH2)n group vibrations band is seen at about 700 cm-1 [17]. The peak can be seen from Fig. 3(bot-tom), but at lower transmittance than that of the biodiesel standard. The ester carbonyl (C=O) group stretching vibration is found between 1500 and 1700 cm-1. The peak appeared at a low intensities in the biodiesel sample produced from Fig. 3(bottom).

3.4. Empirical modelling and influence of operating variables

The biodiesel yield obtained from the outcome of the execution of the two level full factorial designs (23 full-factorial) is shown in Table 3. Experimental results provided in Table 3 were employed in developing a regression model using the general polynomial regression model presented in Eq. (2) as the basis. Several model candidates were explored and Mean Square Error (MSE) was employed as a tool to evaluate the suitable model candidates (see Ref. [16] for details about the use of MSE for model screening). Consequently, the most suitable model candidate was selected. A linear regression model obtained as a suitable candidate model is shown in Eq. (5):

Y = -1.0670(±1.1743) + 1.6233(±0.0018)Г

-0.2088(±0.000)i + 0.9900(±0.0018)W (5)

±1.6145 x 10-11

where T, t and W are the reaction temperature (°C), the reaction time (min) and the catalyst weight (in wt.%), respectively. The experimental yield of biodiesel (ye) was also compared with the predicted biodiesel yield (Ур) as shown in Table 3. The biodiesel yield from the model is in close agreement with experimental values (see Table 3). The points that depicted a strong

University for Science xxx (2015) xxx-xxx Table 4

Cross-validation experiments.

Parameters Biodiesel yield (%)

Expt. run T (°C) t (min) W (wt.%) Experimental Predicted

1 2 40 100 0.2 44.33 43.18

50 120 200 0.8 45.60 55.83

3 60 1.0 54.88 55.56

Table 5

One-way ANOVA analysis of the model.

SS df MS lvalue VR

Between 15.876 1 15.876 28.466 0.558 0.37

Within 113.863 129.739 4

Total 5

MS: mean squares; SS: sum of squares; VR: variance ratio; df: degree of freedom.

deviation were from experimental runs 2, 4 and 8 (see Table 3). Experimental run 4 showed the highest deviation between the predicted and experimental yields. The errors obtained are in a reasonable range because the highest deviation amounted to 13%, indicating that the regression model is suitable to describe the process.

Results from the cross-validation experiment are shown in Table 4. Outcome of the one-way anova analysis conducted on the model using the results from the cross-validation experiment is shown in Table 5.

Cross-validation of the model reveal that VR ^ Fvalue, indicating the validity of the model for predicting the biodiesel yields. Furthermore, Eq. (5) was employed to obtain response surfaces and contour plots to explain the relationship and interactions between the influencing variables and the biodiesel yield. Since it is not possible to represent all the four parameters on a 3-D plot, one variable was held unchanged at a time, and the influence of other two variables on the biodiesel yield is presented on a 3-D surface. Two-dimensional (2-D) plots were employed to explain the interaction between other two variables. The response surface and the contour plots are shown in Figs. 4-6.

A three dimensional response surface and contour plots explaining the influence of catalyst weight (wt.%) and reaction time on biodiesel yield at a constant temperature of 30 °C is presented in Fig. 4(top) and Fig. 4(bottom).

The relationship between the reaction time and biodiesel yield is inversely proportional. This relationship could be attributed to the fact that as the reaction time increased; a reversible reaction occurred resulting in the decrease in biodiesel yield [18]. An increment

M.O. Daramola et al. /Journal ofTaibah University for Science xxx (2015) xxx-xxx

0.5 1 1.5 2 2.5

Catalyst welght(%)

Fig. 4. Response surface plot (top) and contour plot (bottom) of biodiesel yield at 30 °C.

of the biodiesel yield with decreasing reaction time is observed from Fig. 4(top). The reaction time for this current study ranged between 30 and 240 min. A study by Mathiyazhagan and Ganapathi [18] evaluated the effect reaction time on FAME yield. From the study it was observed that a maximum yield was attained within 90 min and beyond that a decline in yield was observed. The transesterification reaction is reversible hence, at longer reaction times there is a reduction in biodiesel due to loss in esters (ester hydrolysis) and soap formation [25]. The relationship between biodiesel yield and reaction time shown in Fig. 3(top) can then be justified by the observation of Mathiyazhagan and Ganapathi [18]. From Fig. 4(top), it could also be deduced that the weight of the catalyst has little effect on the biodiesel yield and this is in agreement with the studies by Omar and Amin [26]. There appears there was no interaction between

0.5 1 1.5 2 2.5

Catalyst weight(%)

Fig. 5. Response surface plot (top) and contour plot (bottom) of biodiesel yield at 30 min.

the reaction time and catalyst weight (see Fig. 4(bot-tom)). Fig. 4(top) and Fig. 4(bottom) further show that the biodiesel yield is only slightly influenced by the catalyst weight. Fig. 4(bottom) also shows that there is no interaction between the catalyst weight and reaction time. Both of these parameters are independent variables and cannot influence each other in any way.

The response surface and contour plots explaining the influence of catalyst weight and reaction temperature on biodiesel yield at constant reaction time of 30 min is shown in Fig. 5.

Again, the catalyst weight showed little effect on the biodiesel yield. As expected, there was an increment in biodiesel yield with increasing reaction temperature. The transesterification of triglycerides has been reported to be an endothermic reaction [27]. According to Le-Chatelier's principle, an increase in reaction temperature

M.O. Daramola et al. /Journal ofTaibah University for Science xxx (2015) xxx-xxx

Fig. 6. Response surface plot (top) and contour plot (bottom) of biodiesel yield at 0.5 wt.%.

for an endothermic reversible reaction favours the forward reaction [28]. No observable interaction between the temperature and catalyst weight was observed (see Fig. 5(bottom)). This observation can also be explained by the collision theory. If the temperature of a substance increases, the rate of the reaction increases as well due to faster moving particles in the reaction vessel. When the rate of successful collisions increases, the conversion of oil to biodiesel, hence increase in the biodiesel yield. Furthermore at increasing catalyst weight, the biodiesel

yield only slightly increased (see Fig. 5(bottom)). In addition, no observable interaction exists between the reaction temperature and the catalyst weight since trans-esterification is endothermic as reported elsewhere [29].

Furthermore, the response surface and contour plots explaining the influence of reaction temperature and reaction time on biodiesel yield at constant catalyst weight of 0.5 wt.% is depicted in Fig. 6.

Fig. 6 depicts the surface response (top) and contour plot (bottom) for the influence of reaction temperature and reaction time at catalyst weight of 0.5 wt.%. The influence of the two variables showed a larger impact on the biodiesel yield. However, Fig. 6(bottom) reveals no interaction between the reaction temperature and time. As explained for the effect of temperature and catalyst weight on the biodiesel yield, increase in biodiesel yield could be attained at higher reaction temperature and time due to the endothermic nature of the trans-esterification of triglycerides [26,27]. These results are in agreement with the reports results of Marchetti and Errazu, where the reaction of free fatty acids and ethanol was carried out using sulfuric acid; and it was shown that biodiesel yield increased with increasing reaction temperature [19]. Fig. 6(top) also shows that there was no interaction between the reaction time and reaction temperature. However, increase in the reaction temperature could increase the rate of the reaction, thereby shortening the reaction time due to the reduction in the viscosity of the oil [18,26].

The performance of calcined sodium silicate for trans-esterification of WCO was compared with other reported solid catalysts as shown in Table 6. When comparing this to a study by Li et al. [9], it is evident that sodium silicate exhibited a higher catalytic performance. Transesteri-fication of WCO to BD catalysed by calcined sodium silicate in this study displayed a higher biodiesel yield at a lower reaction temperature and lower alcohol/oil molar ratio using the same reaction time. A study by Jacobson et al. [30] resulted in a 79% biodiesel yield, which is much higher than the yield obtained in this study. This could be attributed to the higher reaction temperature used in the study. The insignificant effect of the catalyst weight on the biodiesel production, as reported in this

Table 6

Our results compared with literature.

Type of oil Catalyst Catalyst Alcohol Alcohol/oil Reaction Reaction FAME Reference

dosage (wt.%) molar ratio temperature (°C) time (min) yield (%)

WCO Amberlyst-15 5 Methanol 20:1 110 240 30 [9]

WCO ZS/Si 3 Methanol 6:1 200 240 79 [30]

WCO Calcined sodium silicate 2.5 Methanol 6:1 64 240 57.9 This study

M.O. Daramola et al. /Journal ofTaibah University for Science xxx (2015) xxx-xxx

study, is confirmed by the comparison between the catalyst weight used in this study and the one used by Li et al. [9]. In addition, the confirmation from the comparison further strengthens the validity of the linear regression model developed in this study to explain the influence of operating variables on the production of biodiesel from WCO.

4. Conclusions

The use of sodium silicate catalyst for the conversion of WCO to biodiesel was successful, taking into account the FFA content of 3.28% that was above the recommended value of <1%. Sodium silicate was able to convert triglycerides to FAME without the need for esterification pretreatment. The results obtained from the FT-IR and GC/MS show that biodiesel could be produced; especially when the optimum operating conditions are used. The characterization of the sodium silicate catalyst using the chosen methods was successful. It can be concluded that biodiesel yield and reaction temperature have a directly proportional relationship. When transesterification progressed over long periods of time, the biodiesel yield decreased due to the occurrence of a reversible reaction. The catalyst weight has an insignificant effect on the biodiesel yield when the reaction time and temperature are constant. Based on the results obtained from literature, it could be concluded that the developed empirical model could adequately explain the effect of the reaction temperature, reaction time and catalyst weight on the yield of biodiesel from transesterification of WCO over calcined sodium silicate. In addition, the errors between the experimental and the predicted biodiesel yields are within the acceptable limit of 13%. Furthermore, the variance ratio, VR < Fvaiue obtained from the cross-validation experiments indicate perfect agreement of the model output with experimental results and also testifies to the validity and suitability of the model to predict the yield of biodiesel during the transesterification process.

Conflict of interest

There is no conflict of interest References

[1] F. Guo, Z. Fang, Biodiesel production with solid catalysts, in: M. Stoytcheva, G. Montero (Eds.), Biodiesel - Feedstocks and Processing Technologies, InTech, Rijeka, 2011, p. 339.

[2] J. Gerpen, G. Knothe, J. Krahl, Biodiesel production, in: The Biodiesel Handbook, AOCS PRESS, Champaign, 2005.

[3] F. Guo, N. Wei, Z. Xui, Z. Fang, Transesterification mechanism of soybean oil to biodiesel catalyzed by calcined sodium silicate, Fuel 93 (2012) 468-472.

[4] E.F. Aransiola, M.O. Daramola, T.V. Ojumu, M.O. Aremu, S.K. Layokun, B.O. Solomon, Nigerian jatropha curcas oil seeds: prospect for biodiesel production in Nigeria, Int. J. Renew. Energy Res. 2 (2) (2012) 317-325.

[5] E.F. Aransiola, M.O. Daramola, V.T. Ojumu, S.K. Layokun, B.O. Solomon, Homogeneously catalyzed transesterification of Nigerian Jatropha curcas oils to biodiesel: a kinetic study, Modern Res. Catal. (MRC) 2 (2013) 83-89.

[6] Z. Yaakob, M. Mohammad, M. Alherbawi, Z. Alam, K. Sopian, Overview of the production of biodiesel from waste cooking oil, Renew. Sustain. Energy Rev. 18 (2013) 184-193.

[7] A. Talebian-Kiakalaieh, N.A.S. Amin, H. Mazaheri, A review on novel processes of biodiesel production from waste cooking oil, Appl. Energy 104 (2013) 683-710.

[8] Y.Sharma, B. Singh, J. Korstad, Advancements in solid acid catalysts for ecofriendly and economically viable synthesis of biodiesel, Biofuels (2011) 69-92.

[9] M. Li, Y Zheng, Y.Chen, X. Zhu, Biodiesel production from waste cooking oil using a heterogeneous catalyst from pyrolyzed rice husk, Bioresour. Technol. 154 (2014) 346.

[10] A. Gnanaprakasam, V. Sivakumar, A. Surendhar, M. Thiru-marimurugan, T. Kannadasan, Recent strategy of biodiesel production from waste cooking oil and process influencing parameters, Energy (2013) 1-8.

[11] F. Guo, Z. Peng, J. Dai, Z. Xui, Calcined sodium silicate as solid base catalyst for biodiesel production, Fuel Process. Technol. 91 (3) (2010) 322.

[12] F. Guo, C. Zheng, M. Zheng, F. Huang, W. Li, Q. Huang, Soild base catalysts for the production of fatty acid methyl esters, Renew. Energy 53 (2013) 377-383.

[13] M.O. Daramola, D. Nkazi, K. Mtshali, Synthesis and evaluation of catalytic activity of calcined sodium silicate for transesterification of waste cooking oil to biodiesel, Int. J. Renew. Energy Res. 5 (2) (2015)517-523.

[14] S. Geisser, Predictive Inference: An Introduction, Chapman and Hall, London, 1993.

[15] P.A. Devijver, J. Kittler, Pattern Recognition: A Statistical Approach Prentice-Hall, London, 1982.

[16] M.O. Daramola, A.G. Adeogun, Empirical modelling of chemically enhanced backwash during ultrafiltration process, Membr. Water Treat. 2 (4) (2011) 225-237.

[17] S. O'Donnell, I. Demshemino, M. Yahaya, I. Nwandike,L. Okoro, A review on the spectroscopic analysis of biodiesel, Eur. Int. J. Sci. Technol. 2 (7) (2013) 137-146.

[18] M. Mathiyazhagan, A. Ganapathi, Factors affecting biodiesel production, Res. Plant Biol. 1 (2) (2011) 1-5.

[19] M. Mittelbach, P.Tritthart, Diesel fuel derived from vegetable oils, III. Emission tests using methyl esters of used frying oil, J. Am. Oil Chem. Soc. 65 (1988) 1185-1187.

[20] G. Borbely, H.K. Beyer, H.G. Karge, W. Schweiger, A. Brandt, K.-H. Bergk, Chemical characterization, structural features anf thermal behaviour of sodium and hydrogen octosilicate, Clays Clay Miner. 39 (5) (1991) 490-497.

[21] B.C. Smith, Fundamentals of Fourier Transform Infrared Spectroscopy, 2nd ed., CRC Press, Boca Raton, 2011.

[22] N. Noor Hindryawati, G.P.Md. Maniam, R. Karim, K.F. Chong, Transesterification of used cooking oil over alkali metal (Li, Na, K) supported rice husk silica as potential solid base catalyst, Eng. Sci. Technol. Int. J. 17 (2014) 95-103.

10 M.O. Daramola et al. /Journal ofTaibah University for Science xxx (2015) xxx-xxx

[23] M. Tariq, S. Ali, F. Ahmad, M. Ahmad, M. Zafar, Identification, FT-IR, NMR (1H and 13C) and GC/MS studies offatty acid methyl esters in biodiesel from rocket seed oil, Fuel Process. Technol. 92 (3) (2011) 336-341.

[24] M. Bradley, Biodiesel (FAME) Analysis by FT-IR. Available from http://www.thermoscientific.com/content/dam/tfs/ATG/CMD/ CMD%20Documents/Application%20&%20Technical%20 Notes/AN-51258-Biodiesel-FAME-Analysis-by-FT-IR.pdf (accessed 23.10.14).

[25] T. Eevera, K. Rajendran, S. Saradha, Biodiesel production process optimization and characterisation to asses the suitability of the product for varied environmental conditions, Renew. Energy 34 (3) (2009) 762-765.

[26] W.N. Omar, N.A. Amin, Optimization of heterogeneous biodiesel production from waste cooking palm oil via response surface methodology, Biomass Bioenergy 35 (3) (2011) 1329-1338.

[27] R.P. Anitori, Extremophiles: Microbiology and Biotechnology, Caister Academic Press, Norfolk, 2012.

[28] M.A. Fox, J.K. Whitesell, Organic Chemistry, 3rd ed., Jones & Bartlett Learning, Mississauga, 2004.

[29] J.S. Yuan, K.H. Tiller, H. Al-Ahmad, N.R. Stewart, C.N. Stewart Jr., Plants to power: bioenergy to fuel the future, Trends Plant Sci. 13 (2008) 421-429.

[30] K. Jacobson, R. Gopinath, L.C. Meher, A.K. Dalai, Solid acid catalyzed biodiesel production from waste cooking oil, Appl. Catal. B: Environ. 85 (2008) 86-91.