Prediction of cetane numbers of pure biodiesels based on various fuel properties using multiple non-linear correlations

Abstract

Biodiesel has received considerable attention in recent years owing to its advantages over diesel fuel. Density, viscosity, flash point, heating value, and cetane number are among the most critical fuel properties of biodiesel. In the literature, using multiple linear regression, machine learning methods, and group contribution methods, some correlations have been derived to predict the cetane number of pure biodiesels depending on their properties (composition of fatty acid esters, number of carbon atoms, number of double bonds, molecular weight of the fatty acid esters, chain length, saponification number, iodine value, etc.). However, fewer researchers have attempted to establish a correlation between cetane number and other important fuel properties. Therefore, this study aims to employ multiple non-linear regression method for predicting the cetane number of pure biodiesels depending on the density, viscosity, flash point, and heating value. To obtain correlations, experimental data on fuel properties from the literature covering 100 different biodiesels (methyl and ethyl esters) were collected. The prediction performances of the suggested multiple non-linear correlations were compared to the multiple linear correlation frequently suggested in the literature. Reliable correlations with errors lower than 5% and high coefficient of determination values (r2) were derived.