The 27 samples from the experimental design were distributed into training, validation and testing in the proportion of 19 samples (70%), 4 samples (15%) and 4 samples (15%) respectively. The procedures used for the transesterification of the waste soybean oil using methanol and the two forms of catalysts (KOH and NaOH) separately were exhaustively outlined in Ayoola et al.
MINITAB EXPRESS RANDOM DATA SOFTWARE
The software was also used for statistical analysis, which involved the plots of experimental biodiesel yields and statistical modelling. īox-Benkehn BB(4) experimental design (through the application of Minitab 17 software) was used for the design of experiments. Removal of water was done by heating the esterified oil at 110☌ for 20 min. 2 drops of phenolphthalein were added to the mixture and then titrated with 0.1 M KOH until a permanent pale pink colouration was observed. 40 mL of this solution was then added to WSO and heated to 55☌. Impurities were removed through filtration process (using the industrial sieve of 50 μm pore size), FFA was removed through the esterification process which involved mixing 25 mL isopropyl alcohol with 10 mL benzene. Hence, WSO was pre-treated before the transesterification process was carried out.
MINITAB EXPRESS RANDOM DATA FREE
The pre-treatment of WSO involves the removal of impurities, free fatty acid (FFA) and water to prevent low yield of biodiesel (and even formation of other products) during transesterification reaction, ,,. Plots of regression (training, validation, test and overall), mean squared error and error histogram were made using ANN algorithm.ĭepartment of Chemical Engineering, Covenant University, Ota, Nigeria.Ģ. Experimental Design, Materials, and Methods The 27 samples from the Box-Benkehn BB(4) design were distributed into training, validation and testing in the proportion of 19 samples (70%), 4 samples (15%) and 4 samples (15%) respectively. ANN algorithms were sectioned into data division (random), training (Levenberg Marquardt) and performance (mean squared errors). RSM analysis was done through the adoption of three (3) continuous factors: one (1) number of categorical factor, one (1) number of block and one (1) number of replicate for the plots of biodiesel yields and prediction of statistical models, analysis of Sum of Errors coefficients (SE coefficients), p values and F values. The esterification process of free fatty acid (FFA) removal involved the addition of 40 mL of a mixture of 25 mL isopropyl alcohol and 15 mL benzene solution to waste soybean oil (heated to 55☌), as well as the addition of 2 drops of phenolphthalein and the mixture was then titrated with 0.1 M KOH.
Box-Benkehn BB(4) design in Minitab 17 environment was used for the design of experiments, ANN algorithm and RSM software were used for optimization studies. The variables employed in the generation of biodiesel yield data were methanol-oil mole ratio (6 – 12), catalyst concentration (0.7 – 1.7 wt/wt%), reaction temperature (48 – 62☌) and reaction time (50 – 90 minutes). Renewable Energy, Sustainability and the Environment, Engineering, Environmental EngineeringĮxperimental design on transesterification reaction, the use of ANN algorithm, use of Minitab 17 software, model Energy, Engineering, Environmental Science