Glycerol Sensing in Biodiesel Using Turbidimetry

• Published in: Transactions of the ASABE Vol. 52; no. 4; pp. 1261 - 1265
• Main Authors: Zawadzki, A, Shrestha, D.S
• Format: Publication
• Language: English
• Published: 2009
• Subjects: analytical methods; biodiesel; bound glycerol; chemical composition; emulsions; free glycerol; glycerol; mixing; turbidimetry; turbidity; water

Free glycerol and total glycerol are the key quality parameters of biodiesel. Turbidity caused by emulsion formation when biodiesel is mixed with water was investigated as a method to sense free glycerol (FG) and bound glycerol (BG). Turbidity was measured as the absorbance at 600 nm. Six batches of biodiesel samples were spiked with 0%, 0.01%, 0.02%, and 0.03% of FG prepared with glycerol-ethanol solution. A small amount of deionized water was added, and the prepared sample was shaken in a shaker. The turbidity was measured at 3 min intervals at up to 27 min of shaking. Initially, the turbidity increased rapidly with shaking, but it soon leveled off. A linear relationship was found between FG and the square root of turbidity measured after 12 min of shaking (R 2 = 0.87). Although ANOVA showed a significant difference in mean measured turbidities with different levels of FG and BG, a mathematical relationship could not be derived relating BG to turbidity measurement. It is pointed out that the variation in turbidity could have come from unmeasured sterols rather than from BG itself. This method of predicting FG from turbidity was shown to be useful in screening biodiesel batches that need further glycerol removal through additional water washing or other techniques in a biodiesel production process. An absorbance value of less than 0.12 corresponded to an FG value of less than 0.02% with a 95% confidence interval. Four production samples of biodiesel showed that the absorbance limit of 0.12 is adequately conservative because the turbidity of the production samples was found to be slightly higher for the same amount of free glycerol compared to the spiked samples. This observation was ascribed to the presence of other impurities in the production samples, which were maintained at zero level for the control samples.