Performance comparison of empirical model and Particle Swarm Optimization & its boiling point prediction models for waste sunflower oil biodiesel

Abstract

The absence of correlations for predicting the boiling point of biodiesels prevents fuel users to achieve effective engine performance. Among the quality regulator of rudimentary fuel properties is the boiling point and its absence in literature is preventing fuel handlers to achieve actual engine performance. In this study, the mechanism of sunflower oil methanolysis was investigated by Response Surface Methodology (RSM) and Particle Swarm Optimization (PSO). The empirical model (EM) was utilized to correlate the optimal yield and trans-esterification variables for methylic biodiesel production. Thereafter, statistical regression techniques were employed to model the MBP of biodiesel vs. biodiesel fraction and MBP of biodiesel vs. kinematic viscosity. The yield of waste sunflower oil methyl ester (WSOME) (97%) was the uppermost at the methanol/SFO molar ratio of optimal of 6/1, KOH of 1 %wt, and retention time of 78 min. The PSO model exhibited an advanced coefficient of determination, and an inferior value of root mean squared errors related to the RSM model. PSO predicted values, as related to RSM predicted yield shows its dependability and expediency for prediction deprived of conservative experimentation. The fuel properties of the WSOME synthesized were within the ranges of established green fuel standards. The RSM with PSO has been exhibited efficient tools for exploring the methylic biodiesel production from WSO. Least square regression and parabolic equation correlated MBP as a function of bio-diesel fraction and MBP as a function of kinematic viscosity. In conclusion, the results of this study can be useful for biodiesel production from industrial waste oil and the prediction of MBP in the biodiesel industry.