Statistical analysis on prediction of biodiesel properties from its fatty acid composition

The present work deals with the statistical analysis to determine the relationship between the vital biodiesel properties (cetane number, density, viscosity, heating value) and fatty acid compositions of nine different types of biodiesels originated from edible oil non-edible oil, waste oil, and animal oil. Multiple linear regression analysis (MLR) was used to develop the mathematical models to predict the properties from the saturated (lauric, myristic, stearic) and unsaturated fatty (oleic, linoleic, linolenic) acids composition. The developed models were then validated with experimental data from the literature to determine their predictive capability. The models developed for cetane number and density were highly statistical and successfully predicted the respective properties of randomly selected biodiesel from the literature. On the other hand, predictive models for kinematic viscosity and heating value were ineffective; however, the error between experimental and predicted values was sufficiently minimal for heating value.