CONCLUSIONS
The shelf-life of biodiesel is an important property that determines how long it can be stored at low temperatures at fuel terminals and in vehicle tanks and fuel systems. The present work was organized with the objective of developing mathematical equations for estimating the shelf-life of biodiesel at T = 25 °C (298.15 K). To meet this objective, two types of models, Model A and Model B , were obtained from regression analysis where ln(IPR) was expressed as a linear function of T and T−1 (Eqs. 1 and 3). While the experimental IPR data were acquired with measurement temperatures in the range 50-140 °C (323.15-413.15 K), the shelf-life data (SLA and SLB) were calculated by extrapolating the model equations to T = 298.15 K (Eqs. 2 and 4).
Both Model A and Model B type equations demonstrated good correlation between ln(IPR) and T (R² ≥ 0.985) and T−1 (R² ≥ 0.989) and predicting IPAand IPB values within corresponding measurement temperature ranges for the five FAME (CaME, PME, PME, MeC18:1 and MeC18:2) studied in this work. Confirmation analysis of these results showed that calculated IPB data from the Model B equations were more accurate than IPA (fromModel A equations), with respect to measured IPR data. Four of the five FAME exhibited lower MAD and RMSD values for calculated results from the Model B type equations (exception: MeC18:1).
The shelf-life results from extrapolation of the model equations showed that results from Model B (SLB) were consistently greater than results from Model A(SLA) equations. The results for CaME, PME, MeC18:1 and MeC18:2 demonstrated that SLB values exceeded SLA values by an order in magnitude. While Model A is traditionally used to estimate shelf-life data in the fats and oils industry, this work showed that extrapolating Model B type equations appeared to yield more realistic shelf-life estimates for CaME, PME, SME and MeC18:1. The estimated shelf-life results for MeC18:2 (SLA = 74.8 h and SLB = 116 h) seemed too low for realistic storage at T = 298.15 K.
Future studies on this topic might include acquiring more experimental data for Ea and Z0 factors for FAME mixtures under conditions similar to those used to measure IPR at variable temperatures. Acquiring such data would be helpful in the development of kinetic modeling approaches to estimate shelf-life data for biodiesel at low temperatures. For example, an earlier study (Dunn, 2020) suggested that the autocatalysis model f(α) = α½(1 − α) may work well for FAME mixtures that are high in MeC18:1 concentration (Dunn, 2020).