OPTIMIZATION OF BIODIESEL PRODCUTION FROM TRANSESTERIFICATION OF WASTE COOKING OILS USING ALKALINE CATALYSTS M.M. Zamberi 1,2 a, F.N.Ani 1,b and S. N. H. Hassan 2,c 1 Department of Thermodynamics and Fluid Mechanics,Faculty of Mechanical Engineering, Universiti Teknologi Malaysia, 81310 Skudai,Johor, Malaysia 2 Faculty of Mechanical Engineering, Universiti Teknikal Malaysia Melaka, Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia Email: a mahanum@utem.edu.my, b farid@mail.fkm.utm.my, c norhabibah@utem.edu.my
OBJECTIVE Waste vegetable oil(wvo) always being one of the alternative options due to low in price and easy to obtain. The transesterification of WVO with methanol in the presence of potassium hydroxide (KOH) is studied in order to produce biodiesel. The Statistical method using response surface methodology (RSM) based on central composite design (CCD) was being used in this research to assist the optimization process and provide a research strategy in studying the interaction of the parameters and influenced variables.
LITERATURE Several studied have been carried out to optimize the process by using the low price feedstock such as waste vegetable oil (WVO) that can reduce the disposal problem. Massive amount of WVO that generated from fast food restaurant and household, may lead to environmental and health problems if not properly managed (S. Liu et. al., 2010). Due to the higher viscosity content in WVO, it is necessary to convert the WVO into alkyl esters using transesterification process with the aid of alkaline catalyst. The properties of the biodiesel such as viscosity, acid value, sulfur content and heating value are varying with type of feedstock (J.M.Marchetti., 9012) Statistical method using response surface methodology (RSM) based on central composite design (CCD) was being developed to assist the optimization process and provide a research strategy in studying the interaction of the parameters. Variables that influenced the optimization process on conversion of triglycerides such as molar ratio, catalyst weight and reaction time were investigated.
Materials and Methods Transesterification reaction of WVO, purchased from a local restaurant with 480 g methanol with molar ratio 4:1, and KOH alkaline-based catalysts,iscarriedoutintherangeof60±2 C and 25 minute throughout the experiment. After separation and washing with warm deionized water for several times, the excess methanol and water were both eliminated by evaporation under atmospheric condition
Design of Experiments A central composite design (CCD) model was employed which led to 20 experimental runs with 6 axial points, 8 factorial points and 10 center points. Methanol/ oil molar ratio denoted by x 1, (3:1-7:1), catalyst concentration, x 2 (0.3-1.1wt.%)andreactiontime,x 3 (15-75min)withY(FAME%)ischosentobethe response variable. Thereallevels andcoded of eachparameteraregiven in Table1while Table2 shows the results and the design matrix of the experiments. Table 1: Codes, ranges and levels of independent variables (experimental) Coded Level Factor Unit -2-1 0 1 2 Molar ratio (x 1 ) mol/mol 3 4 5 6 7 Weight catalyst (x 2 ) wt% 0.3 0.5 0.7 0.9 1.1 Reaction time (x 3 ) minute 15 30 45 60 75
Design of Experiments Table 2: Central composite design of three independent variables (factors) with experimental response value Run Molar ratio x 1 Factors Level Factors Actual Value FAME (%) Weight catalyst Reaction Molar ratio Weight Reaction Experimental Prediction x 2 time x 3 x 1 catalyst, x 2 time, x 3 1-1 -1-1 4 0.5 30 94.15 94.12 2-1 -1 1 4 0.5 60 95.52 92.76 3-1 1-1 4 0.9 30 90.21 89.49 4-1 1 1 4 0.9 60 83.14 83.32 5 1-1 -1 6 0.5 30 87.82 87.53 6 1-1 1 6 0.5 60 84.20 84.81 7 1 1-1 6 0.9 30 87.32 89.96 8 1 1 1 6 0.9 60 82.51 82.43 9 0 0 0 5 0.7 45 86.12 87.22 10 0 0 0 5 0.7 45 88.56 87.22 11 0 0 0 5 0.7 45 85.98 87.22 12 0 0 0 5 0.7 45 87.22 87.22 13 0 0 0 5 0.7 45 86.97 87.22 14 0 0 0 5 0.7 45 88.12 87.22 15-2 0 0 3 0.7 45 91.38 93.04 16 2 0 0 7 0.7 45 86.99 85.56 17 0-2 0 5 0.3 45 90.18 91.42 18 0 2 0 5 1.1 45 85.43 84.42 19 0 0-2 5 0.7 15 93.05 92.21 20 0 0 2 5 0.7 75 82.23 83.32
Statistical Analysis A double 5- level 3- factor central composite design(ccd) coupled with response surface methodology was used to analyzed the experiments data to fit the following second-order equation (V.G. Shashikant & R.Hifjur., 2006) Analysis of variance (ANOVA) was being selected to perform the experimental data regression and analysis. The coefficient of determination R 2 was also being used to evaluate the performance and accuracy of the above polynomial model.
RESULT AND DISCUSSION Table 3 ANOVA for model terms Source df Sum of squares Mean squares F-value p-value x 1 -Molar ratio 1 56.063 56.0627 18.52 0.002 x 2 -Weight catalyst 1 49.035 49.0350 16.20 0.002 x 3 -Reaction time 1 79.968 79.9683 26.42 0.000 x 1 x 2 1 24.957 24.9571 8.24 0.017 x 1 x 3 1 0.932 0.9316 0.31 0.591 x 2 x 3 1 11.592 11.5921 3.83 0.079 x 2 1 1 6.811 6.8105 2.25 0.165 x 2 2 1 0.774 0.774 0.26 0.624 x 2 3 1 0.453 0.4528 0.15 0.707 R 2 = 0.8835, R 2 adj.= 0.7786, R 2 pred.= 0.2051 Significant at 5% level The conversion of biodiesel obtained from the experiment was ranged from 82.23% up to 95.52% in comparison with predicted values(table 2). Analysisofvariance(ANOVA)oftheexperiment(Table3)indicatedthevaluesforR 2 and adjustedr 2 ;0.8835and0.7786respectivelywhichshowsthatanacceptablelevelforthe model accuracy which indicates the high degree of correlation between experimental and prediction data for FAME. Theeffectofprocessvariables; x 1 (molarratio),x 2 (catalystweight)andx 3 (reactiontime) aswellasthecombinedeffectofparameters x 1 x 2 weredeterminesassignificantmodel terms(p-values<0.05),whereasx 1 x 3,x 2 x 3,x 12,x 22,andx 32 hadinsignificanteffectonthe ester yield(p-values > 0.1).
The final equation in terms of coded factors is outlined in following equation: The above equation indicated that the conversion of WVO into biodiesel has linear and quadratic effects on all the three variables. AsillustratedinTable3,reactiontimehasthelargesteffecton theconversionof biodiesel as it has the highest F-test value(26.42) followed by molar ratio(18.52) and catalyst weight(16.20). The interaction effects between variables especially molar ratio and catalyst weight with F-test value 8.24 also shows some significant interaction.
RESULT AND DISCUSSION Fig. 1: Effects of molar ratio (a), effects of catalyst concentration (b) and effects of reaction time (c) on FAME %. Fig. 1 Effects of molar ratio (a), effects of catalyst concentration (b) and effects of reaction time (c) on FAME %.
RESULT AND DISCUSSION Fig.1(a)shows the effect of methanol to oil molar ratio. Each point shows the mean percentageoffameforallrunsatthesamemethanoltooilmolarratiowhere itwasa slight decreased in the FAME percentage when further increased in the molar ratio of methanoltooilduetothefactthatusinghighamountofalcoholwillleadtodifficultiesin the glycerol separation process(m. Rahimi et. al., 2014) Fig.1 (b) shows the effect of catalyst concentration on the FAME % and each point shows the mean percentage FAME for all runs at the catalyst concentration. AfterhavingthehighestmeanFAMEpercentageof90.42at4wt.%,anincreasingin theamountofkohfrom0.7wt.%to1wt.%ledtodecreaseinoffameto85.43%. Apossible explanation for this might be because of the formation of soap that being develop during the huge excess of catalyst. The effect of reaction time on biodiesel yield is shown in Fig.1(c). The individual effect of reaction time towards biodiesel yield has a negative correlation. The extending of the reaction time in the transesterification process may contribute to the decreased of biodiesel yield and causing more formation of free fatty acids. All the biodiesel was evaluated according to the ASTM standard D 6751 and EN14214.
CONCLUSION The optimum condition from the model equation was obtained at 92.60% with molar ratio of 4:1 mol/mol, 0.5033 wt% catalyst concentration and reaction time of 60 minutes while under the same optimum conditions the experimental biodiesel yield was observed at 95.52%. TheR 2 valuewasdeterminedtobe0.8835whichindicatesthat 88.35% of the model in the second equation explained all the variability of the FAME around its mean. This also increases the accuracy of the predicted value with regard to experimental result. All the biodiesel properties were found within the biodiesel standard limits
THANK YOU