INVESTIGADORES
NIGRO Norberto Marcelo
artículos
Título:
Prediction of the physico-chemical properties of vegetable oils using optimal non-linear polynomials
Autor/es:
ALVISO, DARIO; AGUERRE, HORACIO; NIGRO, NORBERTO; ARTANA, GUILLERMO
Revista:
FUEL
Editorial:
ELSEVIER SCI LTD
Referencias:
Año: 2023 vol. 350
ISSN:
0016-2361
Resumen:
Vegetable oils (VOs) comprise 90%–98% triglycerides (three fatty acids esters and glycerol), with trace amounts of mono-glycerides and di-glycerides. The content of glycerides in VOs can vary depending on the specific type of oil, the processing methods used, and other factors such as the cultivar and harvest date. VOs have been examined for usage in different applications due to their physicochemical properties, including stationary engines, big ships, and Diesel engines of low and medium speed. There are around 350 VOs that have the potential to be used as fuel sources, the vast majority of which have yet to have their physicochemical properties investigated. Regression models based purely on VOs fatty acid content are beneficial in this context. This study conducts a regression analysis of VOs density (DE), kinematic viscosity (KV), flash point (FP), and low and high heating values (LHV and HHV) as a function of their fatty acids. Several experimental databases were selected, including the values of VOs fatty acid composition and physico-chemical properties. Optimal non-linear polynomials were chosen for the regression procedure. Scheffé polynomials offer different fitting alternatives to adjust the VOs experimental databases using their five main fatty acids: from simple linear polynomials (including five terms) to full cubic polynomials (including 35 terms). The polynomials are validated by showing how well their results correspond with the experimental databases. The standard error values for the proposed full polynomials concerning the databases for DE, KV, FP, LHV, and HHV are 0.70%, 7.79%, 7.86%, 1.66%, and 0.19%, respectively.