Journal of Scientific & Industrial Research Vol. 75, March 2016, pp. 188-193 Using Response Surface Methodology in Optimisation of Biodiesel Production via Alkali Catalysed Transesterification of Waste Cooking Oil R Naidoo 1, B Sithole 1,2 * and E Obwaka 1 1 Discipline of Chemical Engineering, University of KwaZulu-Natal, Durban, South Africa. *2 Forestry and Forest Products Research Centre, Natural Resources and the Environment, Council for Scientific and Industrial Research (CSIR), Durban, South Africa. Received 13 April 2015; revised 24 May 2015; accepted 22 November 2015 The report focuses on optimisation of alkali catalysis as a process for producing biodiesel from waste cooking oils. Biodiesel production parameters that were optimised were methanol to oil ratio, catalyst concentration, reaction temperature, and reaction time. A statistical experimental design was conducted using the central composite design method and surface methodology, and the results obtained were analysed using a statistical software package to predict the optimal yields and parameters for the process. The predictions were analysed and the most suitable parameters for biodiesel production were selected. From the results the optimum parameters for biodiesel production were a reaction temperature of 68.4 o C, a reaction time of 1.9 hours, a catalyst concentration of 0.75 wt % potassium hydroxide, and a 0.3:1 methanol to oil weight ratio. The optimum yield of biodiesel from these optimum parameters was predicted to be 98.5%. Thus, alkali catalysis was determined to be a suitable process for production of biodiesel from waste cooking oil. Keywords: biodiesel, waste oil, catalysis, optimisation, transesterification. Introduction Two factors which affect the cost of biodiesel production are feedstocks cost and processing cost. Waste cooking oils and greases are abundant worldwide and are sometimes illegally sold to the poor as new products 1. The materials can be used as low cost feedstock for biodiesel production. This report is concerned with synthesis of biodiesel from a low grade feedstock, waste cooking oil, which does not compete with the food industry. Refined vegetable oils contain glycerides, components that are easily converted to fatty acid methyl esters (FAMEs), which are biodiesel components. The conversion is facilitated by alkali catalysis, a process that may not work efficiently on low grade feedstocks which may contain lower amounts of glycerides and significant amounts of free fatty acids and moisture content. Thus recipes used for biodiesel production from refined vegetable oils need modification for application on low grade feedstocks such as waste cooking oils and crude tall oil from the pulp and paper industry. A recent report has shown that biodiesel can be efficiently made from crude tall via use of expensive Author for correspondence E-mail: bsithole@csir.co.za catalysts 2. This report focuses on optimisation of alkali catalysis as a process for producing biodiesel from waste cooking oils. Biodiesel production parameters that were optimised were methanol to oil ratio, catalyst concentration, reaction temperature, and reaction time. A statistical experimental design was conducted using the central composite design method and surface methodology, and the results obtained were analysed using a statistical software package to predict the optimal yields and parameters for the process. The predictions were analysed and the most suitable parameters for biodiesel production were selected. Experimental Waste cooking oil A sample was collected from a fast-food restaurant in Durban, South Africa. The oil was sieved to remove any particulate matter and then boiled for 30 minutes at 110 o C to remove water in the oil. Catalyst Transesterification using alkali catalysis was used according to reports 3-5. The methoxide catalyst was obtained by dissolving potassium hydroxide flakes in methanol.
SITHOLE et al.: USING RESPONSE SURFACE METHODOLOGY IN OPTIMISATION OF BIODIESEL PRODUCTION 189 Biodiesel synthesis parameters The parameters that were studied in the synthesis of biodiesel were methanol to oil weight ratios, catalyst concentration, reaction temperature, and reaction time To avoid unwanted side-reactions, the following conditions gleaned from the literature were studied: reaction temperature between 30-60 o C reaction time between 60-120 minutes catalyst concentration between 0.5-2% methanol to oil weight ratio between 0.3:1 0.5:1 These conditions were reported to achieve yields of 90-98% percent 3-5. Statistic Experimental Design and Analysis Statistical experimental design forms an essential part of any laboratory work as it limits the number of time consuming experiments and also ensures that most suitable data can be obtained from minimum experimental work 5.The central composite design (CCD) was selected for use in experimental analysis, in order to evaluate the effect of parameters individually on the yield of FAMEs. A linear response was required, which is fairly simple but when evaluating effects of parameters simultaneously on the yield of FAMEs a second order response was required. The CCD is a design widely used for estimating second order response surfaces. It is perhaps the most popular class of second order designs 5. In the cube each plane signifies a parameter being evaluated. For example, x-plane signifies temperature and y-plane signifies methanol to oil ratio. The levels +1,-1,0 signify conditions for each parameter, e.g., +1 in x-plane and +1 in the y-plane simply imply maximum conditions for temperature and methanol to oil ratio. Coordinates on the cube are combinations of each parameter and condition. A closer examination of the cube reveals points protruding out of the cube - these are rotatable design points used to test F-beyond conditions. The rotatable design point coordinates are calculated as follows: W = (2 k ) 0.25 (1) where: k = number of factors or parameters; W = rotatable design axial point In order to perform the experiments, centre and axial points were selected as well as the level of investigation of each point. Thus the parameters selected were: 16 cube, 2 centre, and 8 axial points. The full experimental design is shown in Table 1. To obtain experimental points, a combination of coordinates needs to be taken. This is a rigorous procedure that was simplified by using a simple statistical software package known as Design-Expert 8 (Statease). Contour plots were produced from experimental data, analysed by Design Expert 8 to yield optimum conditions, and a quadratic equation was derived 5. Biodiesel synthesis via transesterification Alkali transesterification was selected due to its extensive use in industry as suggested in the literature review. Another reason for selecting alkali transesterification was the assumption that used cooking oil was chemically similar to virgin cooking oil which implies that it also has low free fatty acid content 6. Alkali transesterification can be carried with either sodium hydroxide or potassium hydroxide as suggested in the literature review. Experimental work was carried out using potassium hydroxide as the preferred choice. A schematic of the transesterification process is shown in Figure 1.The catalyst was mixed with the oil in a beaker, placed in a water bath and left to react at a constant suitable temperature. The solution within the beaker was stirred every five minutes to ensure uniform distribution of reactants.the reaction temperature was varied according to synthesis parameters and experimental trial and error from 35 to 75 0 C whereas the reaction time was from 0.4 to 3.4 hours. Purification of the biodiesel After transesterification the biodiesel contains soap, glycerol and methanol 7. These substances are considered impurities for the following reasons 7 : Soap makes biodiesel viscous resulting in plugging of filters in vehicle engines; Glycerol is a natural disinfectant and causes residue in vehicle diesel tanks to wash off and mix with the fuel - this also results in clogging of vehicle filters; Methanol lowers the flash point of biodiesel, 1% of methanol can lower the flash point of biodiesel from 170 o C to 40 o C. When biodiesel has a low flash point it becomes dangerous to transport and handle. The impurities can be removed by washing the biodiesel water since they are soluble in water 7. If need be, glycerol can be separated by density since it is not miscible with biodiesel. However, no significant amounts of glycerol are expected from the low-grade waste cooking oil.thus, after each
190 J SCI IND RES VOL 75 MARCH 2016 Table 1 Alkali Catalysis Experimental Points Levels Variable Units -w -1 0 +1 +w Time Hours 0.4 1.2 1.9 2.7 3.4 Methanol :Oil Ratio 0.1:1 0.2:1 0.3:1 0.4:1 0.5:1 Catalyst Concentration wt% 0.25 0.50 0.75 1.00 1.25 Temperature C 35 45 55 65 75 Run no. Time Methanol: Oil Catalyst Conc. Temperature Response(%FAME) 1-1 -1-1 -1 56.58 2 1-1 -1-1 94.22 3-1 1-1 -1 94.72 4 1 1-1 -1 85.3 5-1 -1 1-1 96.96 6 1-1 1-1 85.7 7-1 1 1-1 95.97 8 1 1 1-1 79.48 9-1 -1-1 1 96.57 10 1-1 -1 1 86.71 11-1 1-1 1 92.64 12 1 1-1 1 90.36 13-1 -1 1 1 91.71 14 1-1 1 1 90.67 15-1 1 1 1 88.12 16 1 1 1 1 86.45 17-2 0 0 0 79.31 18 2 0 0 0 83.1 19 0-2 0 0 95.38 20 0 2 0 0 62.27 21 0 0-2 0 85.4 22 0 0 2 0 49.1 23 0 0 0-2 93.06 24 0 0 0 2 98.55 25 0 0 0 0 97.06 26 0 0 0 0 87.33 Fig. 1 Schematic of the transesterification process. transesterification experiment the mixture was cooled in water to prevent the reaction from proceeding any further. It was then left to settle for two hours after which it was transferred to a separatory funnel to separate the layers of biodiesel and glycerol. Glycerol has a higher density than biodiesel and was removed by drainage via the bottom of the funnel. The biodiesel was then purified by successive washings with 250 ml of water at 50 0 C until a clear water effluent was achieved. Washing was performed to remove impurities, viz, residual glycerol, soaps and catalyst. Care was taken not to stir or shake the biodiesel with water vigorously as this causes saponification. Assessment of biodiesel yield The yield of biodiesel from each experiment was determined by quantifying the loss of glycerides in the oil after transesterification. Assessment of biodiesel quality The quality of the biodiesel produced was ascertained by measurement of FAMEs in the samples using standard gas chromatographic procedures.
SITHOLE et al.: USING RESPONSE SURFACE METHODOLOGY IN OPTIMISATION OF BIODIESEL PRODUCTION 191 Results and Discussions Quality of biodiesel Analysis of the biodiesel samples by GC/MS reveal that the biodiesel was of high quality and was comprised of FAME compounds that are typically found in biodiesel produced from high grade refine vegetable oils. Statistical data The software package Design Expert was used for statistical data manipulation and analysis. The software was used to generate the combination of statistic levels for parameters. The responses and the combinations of statistic levels were used to generate contour plots which indicate how changes in various parameters affect the response. The programme uses the sum of squares method to compute constants for a quadratic equation that relates parameters or process variables to the yield of FAMEs. Model adequacy checking was done by observing residual plots and an ANOVA table. P-values on the ANOVA table describe the significance of each factor thus the quadratic model was reduced and the contribution of each factor was determined. Design Expert performed successive substitution on the quadratic equation to yield maximum and minima points on the quadratic curve. Using these maximum and minima points the most suitable conditions for optimum yield of FAMEs was predicted. When using the optimisation function on Design Expert it should be noted that optimum statistical levels are obtained. These levels are then converted to actual parameter values via interpolation.table 2 illustrates the contribution of each experimental factor: the variables A, B, C, and D Table 2 Contribution of experimental factors Factor Contribution (%) A-Time 7.76 B-M:O 54.04 C-Catalyst Concentration 55.97 D-Temperature 47.93 AB 56.69 AC 57.86 AD 21.15 BC 55.32 BD 40.02 CD 47.54 A 2 33.83 B 2 45.73 C 2 83.72 D 2 46.78 represent Reaction Time, Methanol to Oil weight ratio, Catalyst Concentration and Reaction Temperature, respectively. The terms AD, AC etc. are interaction terms. Interaction terms simply show the combined effects of two factors on the yield of biodiesel for example the term AD shows the effect of Reaction Time and Temperature on Biodiesel yield (%FAMEs). %Biodiesel = 92.195-2.18B-2.28C-2.83AD-2.91AC- 4.99C 2 (2) Equation (2), which is a quadratic equation, was obtained via non-linear regression. The quadratic equation was reduced to only include terms that significantly affect the yield of biodiesel. Catalyst concentration had the highest effect on the yield followed by the combined effect of reaction time and catalyst concentration. Contour Curves The contour curves were constructed from experimental points. They describe trends that occur when various conditions are changed. In this case the contour curves describe the trend in yield of FAMEs when Reaction Time, Catalyst Concentration, Methanol to oil ratio, and Reaction Temperature are decreased or increased.figure 2 shows that increase in reaction time results in increase in yield of FAMEs but high reaction times lead to loss of methanol via evaporation. However, as illustrated, excess amounts of methanol cause a decrease in biodiesel yield.long reaction time and excess catalyst result in decreased yield of biodiesel. However, short reaction times and low catalyst also result in low biodiesel yield. The optimum point on the contour curve (not shown) is clearly at midpoint catalyst and reaction time where the yield is 92.44%.Figure 3 indicates that higher Fig.2 Effect of Reaction Temperature and Methanol to Oil ratio on yield of FAMEs
192 J SCI IND RES VOL 75 MARCH 2016 up.from the results above the optimum parameters for biodiesel production were a reaction temperature of 68.4 o C, a reaction time of 1.9 hours, a catalyst concentration of 0.75wt% potassium hydroxide, and a 0.3:1 methanol to oil weight ratio. The optimum yield of biodiesel from these optimum parameters was predicted to be 98.5% - this value is very close to the highest yield of 98% reported in the literature 8. Fig. 3 Effect of Reaction Time and Reaction Temperature on yield of FAMEs. temperatures lead to increased yield of biodiesel. However, some contour curves exceed 100% yield which is within experimental error. Temperatures above 70 o C cause methanol to vaporise which lowers the biodiesel yield. Therefore, temperatures below 70 o C are essential for high yields of biodiesel with a midpoint reaction time of 1.9 hours. Overall, the trend in the contour curves shows that increasing temperature and reaction time result in increased biodiesel yield.the data showed that the ideal conditions for a high biodiesel yield are when methanol to oil weight ratio and catalyst concentration are at about midpoint on a statistical level. The trend illustrated by the contour curves show clearly that increasing catalyst concentration and methanol ratio increases biodiesel yield.analysis of the relevant contour plot (not shown) illustrates that a midpoint of methanol to oil ratio and a high reaction temperature results in a high yield of biodiesel. The trend of the contour curves illustrate that an increasing reaction temperature results in an increased yield of biodiesel.increase in catalyst concentration increased yield of FAMEs till the midpoint after which the yield decreased. As ascertained in the contour curves (not shown), reaction temperature should be kept high enough for the reaction rate to be rapid but low enough to avoid boiling off the methanol. Therefore, a reaction temperature of 68.4 o C was deemed optimum. All the contour plots above showed twisting curves which implied that there were indeed interactions between parameters. The contour plots also have convergence points which illustrate that optimum conditions for the production of biodiesel were investigated thoroughly in the experimental design set Conclusions From the quadratic equation (2) it was noted that catalyst concentration and reaction time had the most significant effect on the yield. The following conclusions were drawn from analysing the contour plots: Methanol to oil ratios above 0.4:1 resulted in decreased biodiesel yield. Reaction times above 2.7 hours decreased biodiesel yield. Reaction Temperatures above 65 o C increased biodiesel yield. Catalyst concentrations above 1 wt% decreased biodiesel yield Temperatures above 70 o C were deemed not suitable due to methanol boil off and reaction times above 2.7 hours also resulted in methanol boil off. Methanol boil off resulted in decreased biodiesel yield. Twisting contour curves on plots indicate that interactions between parameters were present during experimentation. From the results above the optimum parameters for biodiesel production were a reaction temperature of 68.4 o C, a reaction time of 1.9 hours, a catalyst concentration of 0.75wt% potassium hydroxide, and a 0.3:1 methanol to oil weight ratio. The optimum yield of biodiesel from these optimum parameters was predicted to be 98.5%. Thus, alkali catalysis was determined to be a suitable process for production of biodiesel from waste cooking oil. References 1 Komintarachat C & Chuepeng S, Solid acid catalyst for biodiesel production from waste used cooking oils, Ind Eng Chem Res, 48 (2009) 9350-9353. 2 Mkhize N M, Sithole B B & Ntunka G, Heterogeneous acid catalysed biodiesel production from crude tall oil: a low grade and less expensive feedstock, J Wood Chem Technol, 35 (2015) 374 385. 3 Demirbas A, Biodiesel from waste cooking oil via base catalytic and supercritical methanol transesterification, Energy Convers Manage, 50 (2009) 923-927.
SITHOLE et al.: USING RESPONSE SURFACE METHODOLOGY IN OPTIMISATION OF BIODIESEL PRODUCTION 193 4 Gokhan C & Selim K, Increased yields in biodiesel production from used cooking oils by a two-step process: comparison with one step process by using TGA, Fuel Process Technol, 89 (2008) 118-122. 5 Bautista L V & Pacheco M, Optimisation of FAME production from waste cooking oil for biodiesel use, Biomass Bioenergy, 33 (2009) 862-872. 6 Schuchardt U, Sercheli R & Vargas R, Transesterification of vegetable oil: a review, J Brazil Chem Soc, 9 (1998) 199-210. 7 Sithole B B, Characterization of biodiesel, Internal Report, Pulp and Paper Research Institute of Canada, Pointe-Claire, Quebec, Canada, 2009. 8 Phan A & Phan T, Biodiesel production from waste cooking oils, Fuel, 87 (2008) 3490-3496.