A Multivariate Regression Model for Estimation of Biodiesel Production from Jatropha Seeds

This paper presents a statistical technique to characterize the biodiesel production process from Jatropha seeds and predict the conversion efficiency of the oil into biodiesel. A multivariate regression model is proposed herein to simultaneously capture the joint effect of the variable operating conditions. It is a general technique and can be used with different production methods, but this paper uses experimental results of the in- situ heterogeneous extraction and transesterification process. The results imply the success of the introduced technique to assess how much the operating variables affect the conversion. Also, the joint effect of the time, catalyst dose, methanol to oil ratio, and hexane to oil ratio has been defined. Experimental measurements have been used to perform this analysis. A linear multiple regression model with coefficient of determination (R 2 ) of 0.999 has been developed whereby the four investigated operation variables are related with the conversion into biodiesel has been developed. Hence, utilizing the coefficients of the developed model have been used to rank their significance, which is the most important variable and to what extent. Accordingly the most important factors are the time and the hexane ratio.