Prediction of the Octane Number: A Bayesian Pseudo-Component Method

Abstract

The energy transition leads to the development of unconventional liquid fuels. Unconventional liquid fuels are produced at a small scale so they are produced with a limited budget and they must be characterized at a cheap price. When liquid fuels are burned in piston engines, they are characterized by the Research Octane Number (RON) and the Motor Octane Number (MON). As the measurement of the RON and the MON is expensive, a cheaper alternative, like the pseudo-component method, is sought. Nevertheless, this method was only developed for the RON, it is not applicable for complex fuels with olen and oxygenates, and its uncertainty has not been characterized. Moreover, it does not dierentiate the isomers. For instance, the iso-parans 1 are considered as a blend of 2-methyl-alkane, 3-methyl-alkane, 2,2-dimethyl-alkane and 2,3-dimethyl-alkane in equal proportions. The authors address the limitations of the pseudo-component method using a Bayesian approach. The validity of the method is demonstrated for three gasoline blendstocks mixed with ve oxygenated molecules: 1propanol, 2-propanol, 1-butanol, 2-butanol and 2-methyl-1-propanol. As a result, the octane numbers are predicted within the theoretical uncertainty bounds and with less than 2% of error.

Predicting Octane Number of Petroleum-Derived Gasoline Fuels from MIR Spectra, GC-MS, and Routine Test Data

“…Regarding the analytical methods used in ON prediction, most studies have employed multivariate methods such as PLSR (partial least square regression) or PCR (principal component regression). However, the non-linear nature of the ON behavior in complex mixtures has led many researchers to build correlations via machine-learning methods or other novel approaches such as artificial neural networks (ANNs), support vector machines (SVMs), Bayesian estimation, and random forest [21,28].…”