Estimation of hydrocarbon types in light gas oils and diesel fuels by ultraviolet absorption spectroscopy and multivariate calibration

Ultraviolet absorption spectra are used in conjunction with multivariate ca libration methods to predict the percent content of saturates (61-99%), mon oaromatics (1-34%), diaromatics (0-5%), and polyaromatics (0-1%) in light g as oil and diesel fuel samples. A total of 114 samples taken from three pil ot plant studies between 1992 and 1994 were used to develop the calibration models. Supercritical fluid chromatography with flame ionization detection was employed as the reference method for quantifying hydrocarbon types. Se veral multivariate calibration methods (classical least squares, multiple l inear regression, principal components regression (PCR), and partial least squares (PLS)) were examined. PCR and PLS gave the best overall performance , with root-mean-squared errors of prediction (absolute) of 0.9% (saturates ), 0.8% (monoaromatics), 0.28 (diaromatics), and 0.05% (polyaromatics) base d on 44 prediction samples. Relative errors (absolute x 100/range) for thes e same species were 2.4%, 2.4%, 3.2%, and 5.3%.