Oxidative stability of biodiesel by mixture design and a four-component diagram

• Published in: Fuel (Guildford) Vol. 219; pp. 389 - 398
• Main Authors: Viegas, Isabelle M.A., Barradas Filho, Alex O., Marques, Edmar P., Pereira, Claudete F., Marques, Aldaléa L.B.
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.05.2018; Elsevier BV
• Subjects: Biodiesel; Biodiesel fuels; Biofuels; Calibration; Corn; Correlation coefficient; Correlation coefficients; Diesel; Mixture design; Oxidation; Oxidative stability; Polynomial models; Polynomials; Quaternary mixtures; Rapeseed; Raw materials; Root-mean-square errors; Soybeans; Stability; Statistical analysis; Statistical methods; Quaternary mixtures; Polynomial models; Oxidative stability; Mixture design; Biodiesel
• ISSN: 0016-2361; 1873-7153
• DOI: 10.1016/j.fuel.2018.01.124

[Display omitted] •A new way to represent four-component diagrams.•Performance of a study on oxidative stability of binary, ternary and quaternary mixtures of biodiesels.•Comparison of polynomial equations up to fourth degree for quaternary mixtures.•Contour maps with a large range of oxidative stability values of biodiesel. Some biodiesels originating mainly from unsaturated feedstocks such as soybean, rapeseed, corn and sunflower do not reach the minimum oxidative stability of 8 h, as specified by standard EN 14214. An alternative to work with this type of biodiesels is to prepare their mixture with others biodiesels that are mainly saturated. This work presents a study employing binary, ternary and quaternary mixtures of biodiesels from soybean, corn, babassu and palm, aiming to improve the oxidative stability by using mixture design and polynomial modeling. For this, 71 mixtures were designed by simplex-lattice and simplex-centroid designs and divided into a calibration set (50 samples) and an external validation set (21 samples) employing the partitioning method based on joint X-y distances (algorithm SPXY). Six polynomials for quaternary mixtures models were built and statistically compared. The quadratic equation achieved the best results, with correlation coefficients (r) 0.9879 and 0.9569 for calibration and validation steps, respectively. Root mean square errors (RMSE) were 0.67 h for calibration and 0.71 h for validation and mean absolute percentage errors (MAPE) were 5.07% and 7.54% for calibration and validation, respectively. The quadratic equation was used to plot the contour maps for the oxidative stability resulting in a new way to represent four-component diagrams. In addition, a variety of possible biodiesel mixtures (binary, ternary and quaternary) are presented from babassu, soybean, corn and/or palm that meet any oxidative stability specifications within the range of 5.08 h–20.88 h, depending on the availability of the type and amount of the feedstock and target oxidative stability.