Comparison of Empirical Equations and Artificial Neural Network Results in Terms of Kinematic Viscosity Prediction of Fuels Based on Hazelnut Oil Methyl Ester


ENVIRONMENTAL PROGRESS & SUSTAINABLE ENERGY, vol.35, no.6, pp.1827-1841, 2016 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 35 Issue: 6
  • Publication Date: 2016
  • Doi Number: 10.1002/ep.12410
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.1827-1841
  • Yozgat Bozok University Affiliated: Yes


This study investigates the prediction of kinematic viscosity values of hazelnut oil methyl ester (HOME) using empirical equations and artificial neural network (ANN) methods under varying temperature and blend ratio conditions with ultimate euro diesel (UED) fuel. Four different fuel blends (20, 40, 60, and 80% by volume mixing ratio) were studied along with UED fuel and pure biodiesel. Tests for kinematic viscosity were performed in the temperature range of 293.15-373.15 K at the intervals of 1 K for each fuel sample. Moreover, physicochemical properties of hazelnut crude oil (HCO), HOME and its blends, and also fatty acid composition of HCO and HOME were measured and discussed in light of ASTM and EN standards. Regression analyses were conducted using MATLAB software to determine the coefficient of determination (R-2), root mean square error (RMSE), and correlation constants. The best R-2 and RMSE values were obtained by Eq. 6 as 0.9999 and 0.0068, respectively. In the analyses conducted using ANN, R-2, and RMSE were obtained as 0.999986 and 0.00149 respectively based on the overall HOME-UED fuel blends. Although R-2 values obtained by these two methods were close to each other, RMSE obtained using ANN was smaller than that of the one obtained by Eq. 6. In conclusion, the ANN method captures the best accuracy for the prediction of biodiesel kinematic viscosity with the highest R-2 of 0.999986 and the lowest RMSE of 0.00149, which is within +/- 1% error range of the experimental data. (C) 2016 American Institute of Chemical Engineers Environ Prog, 35: 1827-1841, 2016