Predicting the viscosity of diesel/biodiesel blends

- Test of 12 models for predicting viscosity of petroleum/biodiesel blends.
- The majority of existing models did not perform well on oxygenates.
- Six models were modified and 3 new were developed.
- All modified and developed models produced R2 ≥ 99.6%.

Viscosity is one of the most important properties of biodiesel and conventional diesel fuels derived from petroleum. A large number of empirical correlations exist in the literature, which predict the blended viscosity of binary oil mixtures but they have not been proved for oil/oxygenate blends. In this work, twelve mixing rules, which were developed for predicting the viscosity of petroleum based fuels, were tested for their predicting accuracy in oxygenated blends. In order to test these models in diesel/biodiesel blends with adequate experimental data, three different diesel fuels and seven biodiesels were blended in eleven different volume fractions, giving 231 samples, whose viscosities were experimentally measured. The majority of the mixing rules tested, produced similar results and predicted the viscosities with poor accuracy, with just one of them exhibiting satisfactory accuracy. In an attempt to improve the predictions of these equations, some of the mixing rules were modified by using alternative forms or new constants. The modified equations gave reasonable predictions for the viscosities of the oxygenated mixtures. Two new equations were developed to estimate the constant in Lederer's model, to avoid the necessity of further experimental data that the original model required for the prediction of the viscosity. The results were in closer agreement with the experimental data and the accuracy was of the same order of magnitude for all the modified equations, R2 ≥ 99.6%. The most accurate estimations were given by the Modified Shu and Barrufet & Setiadarma's mixing rules, and one more rule which was developed in this study based on Grunberg and Nissan's model.