Prediction of Kinematic Viscosity and Density of Biodiesel Produced from Waste Sunflower and Canola Oils Using ANN and RSM: Comparative Study

Abstract

In this paper, the properties of two biodiesel obtained from waste cooking sunflower (WCSME) and waste cooking canola (WCCME) oils and their blends in the temperature range 20–300 °C is measured experimentally. This work focused on the application of Artificial Neural Network (ANN) as predictive tools for prediction the kinematic viscosity and density of the biodiesel. In the present study, temperature, the composition of methyl esters (wt%/100), the number of carbon atoms (NC), the number of hydrogen atoms (NH), molecular weight in g/mol, the number of double bond in the fatty acid chain and volume fraction of WCSME were used as input for the models. Moreover, Response surface methodology (RSM) was used to predict the effects of either temperature and volume fraction or volume fraction only on the selecting biodiesel properties using Two-variable or single variable model, respectively. Consequently, it was found that the single variable has more significant for predicting the kinematic viscosity and density compared to the two-variable model. Accordingly, the results indicated that the proposed ANN approach is able to provide a good agreement with the experimental data with the overall R² of 0.999 compared with the RSM models.