Towards estimating surface tension of biodiesels: Application to thermodynamic and intelligent modeling

Abstract

Recently, increasing demands for energy accelerates the study on renewable energy resources so biodiesels become one of interesting topics for the researchers. Due to wide and user-friend applications of artificial intelligence methods, in this study, an artificial intelligence method based on support vector machine algorithm optimized by Grey wolf optimization algorithm is suggested to estimate the surface tension of biodiesels. To this end, the experimental surface tension dataset has been collected and divided into two datasets of 59 and 19 points for training and testing randomly. After various comparisons with the real surface tension dataset for the proposed artificial intelligence method, three existing models including UNIFAC, Kay and Dalton models have been participated in the comparisons. The determined R-squared values for Kay, Dalton, UNIFAC, and support vector machine are 0.627, 0.6462, 0.8483, and 0.9905, respectively. According to these results, developed model is the best predictive tool for calculation of surface tension of biodiesels. Additionally, the accuracy of biodiesel surface tension databank has been investigated. On the other hand, the impacts of contributed variables in the models on surface tension of biodiesel fuels have been investigated as an another novel point. It explains that the heaviest fractions have been known as the most effective variables on determination of surface tension of biodiesels. Therefore, this study involves a novel and accurate tool for prediction of surface tension of biodiesels and also sensitivity analysis on effective parameters to help researchers in production of cleaner fuels.