Modeling and optimization of sunflower oil methanolysis over quicklime bits in a packed bed tubular reactor using the response surface methodology

The effect of the residence time (i.e. liquid flow rate through the reactor), methanol-to-oil molar ratio and reaction temperature on the fatty acid methyl esters (FAMEs) content at the output of a continuous packed bed tubular reactor was modeled by the response surface methodology (RSM) combined with the 3³ full factorial design (FFD) with replication or the Box-Behnken design (BBD) with five center points. The methanolysis of sunflower oil was carried out at the residence time of 1.0, 1.5 and 2.0h, the methanol-to-oil molar ratios of 6:1, 12:1 and 18:1 and the reaction temperature of 40, 50 and 60°C under the atmospheric pressure. Based on the used experimental designs, the model equations containing only linear and two-factor interaction terms were developed for predicting the FAME content, which were validated through the use of the unseen data. Applying the analysis of variance (ANOVA), all three factors were shown to have a significant influence on the FAME content. Acceptable statistical predictability and accuracy resulted from both designs since the values of the coefficient of determination were close to unity while the values of the mean relative percentage deviation were relatively low (<±10%). In addition, both designs predicted the maximum FAME content of above 99%, which agreed closely with the actual FAME content (98.8%). The same optimal reaction temperature (60°C) and residence time (2.0h) were determined by both designs while the BBD model suggested a slightly lower methanol-to-oil molar ratio (12.2:1) than the 3³ FFD model (12.8:1). Since the BBD realization involved three times smaller number of experimental runs, thus requiring lower costs, less labor and shorter time than the 3³ FFD, it could be recommended for the optimization of biodiesel production processes.