Determination of Octane Numbers and Reid Vapor Pressure of Commercial Petroleum Fuels Using FT-Raman Spectroscopy and Partial Least-Squares Regression Analysis

Abstract

A Fourier transform Raman spectrometer was used to collect the Raman spectra of 208 commercial petroleum fuels. The individual motor and research octane numbers (MON and RON, respectively) were determined experimentally using the industry standard ASTM knock engine method. Partial least-squares regression analysis was used to build regression models which correlate the Raman spectra of 175 of the fuels with the experimentally determined values for MON, RON, and pump octane number (the average of MON and RON) of the fuels. Each of the models was validated using leave-one-out validation. The standard errors of validation are 0.415, 0.535, and 0.410 octane units for MON, RON, and pump octane number, respectively. By comparing the standard error of validation to the standard deviation for the experimentally determined octane numbers, it is evident that the accuracy of the Raman determined values is limited by the accuracy of the training set used in creating the models. The Raman regression models were used to predict the octane numbers for the fuels which were not used to build the models. The results compare favorably with the leave-one-outvalidation. Also, it is demonstrated that the experimentally determined Reid vapor pressures are highly correlated with the Raman spectra of the fuel samples and can be predicted with a standard error of 0.568 psi.