Evaluation of mixing rules to predict viscosity of petrodiesel and biodiesel blends

Abstract In the face of lacking an equation to predict important characteristics of new biofuels, thirty mixing rules originally developed to calculate the viscosity of petroleum and its fractions were tested, using 303 experimental data of biodiesel and petrodiesel binary blends from literature, covering a variety of vegetal oils and types of petrodiesel. The initial criterion used to evaluate the accuracy of mixing rules to predict viscosity of blends was relative standard error (RSE). Seven of the thirty rules exhibited RSE higher than 25%, and only nine exhibited relative standard error lower than 5% (e.g., excellent agreement between observed and estimated values). For these nine rules, further statistical analyses were applied to strengthen the evaluation. The results revealed that the Lederer equation exhibited the best performance to predict viscosity of binary blends of biodiesel with petrodiesel, with estimation error below 4.29% and residual balance (-7) indicating no tendency of over- or under-estimate values.