Measurement and Prediction of Density and Viscosity of Different Diesel-Vegetable Oil Binary Blends

Abstract

Abstract Vegetable oils can be considered as an alternative or emergency fuel for diesel engine. However, vegetable oils result in operational and durability problems for the long-term operation because of being much more viscous than diesel fuel. To eliminate this drawback, blending of vegetable oils with diesel fuel or alcohol is one of the most widely used techniques. In the existing literature, many studies are available on the measurement and prediction of density and viscosity of binary blends (especially biodiesel (BD)-diesel fuel (DF) blends), although, there is still the lack of comprehensive studies in which reliable density and viscosity data are presented, new regression models are proposed and compared with other regression models for waste cooking oil (WCO)-DF binary blends. Therefore, in the present study, (1) WCO was blended with DF on the volume basis of 2, 4, 6, 8, 10, 15 and 20 %, (2) the measurements of viscosities and densities of the binary blends were performed at various temperatures (278.15–343.15 K) in accordance with DIN 53015 and ISO 4787 standards, respectively, (3) the variations of viscosity and density values of binary blends vs. temperature were evaluated, (4) the new rational and exponential models as a function of temperature were fitted to the experimental data measured by the authors and Baroutian et al. (regarded as typically different data), and finally (5) the models were also compared to Yoon et al. and linear models, previously proposed by other authors, in order to investigate their reliability. According to results, (i) the best correlation was obtained by the rational model with the lowest maximum relative errors of 2.9679 % and 3.2725 % for the viscosity data measured by the authors (WCO-DF blends) and Baroutian et al. (palm oil (PO)-DF blends), and (ii) for the density data of WCO-DF and PO-DF binary blends, the best correlation was obtained using the exponential model giving the lowest maximum relative errors of 0.0470 % and 0.0581 %, respectively.