Estimation of biodiesel cetane number, density, kinematic viscosity and heating values from its fatty acid weight composition

A detailed statistical investigation was conducted in order to correlate some important biodiesel properties with the respective methyl ester weight composition in fatty acids. The examined properties were cetane number, density, kinematic viscosity and heating values (lower and higher). The chosen method for the correlation was the multiple linear regression analysis. A comprehensive data set was chosen for each interesting property as the basis for the formulation of the linear relations with respect to the eight, most important, fatty acids (lauric, myristic, palmitic, stearic, palmitoleic, oleic, linoleic, linolenic). The derived correlations were then verified against other experimental data, selected from various sources, with the aim to assess their predictive capability. It was found that for both the cetane number and density, the derived correlations were, on the one hand highly statistical, and, on the other, proved successful in predicting the corresponding properties from randomly selected samples reported in the literature. On the contrary, for both heating values and kinematic viscosity, the statistical correlations were weaker, although for the HHV the relative error between experimental and predicted values was adequately small.