REFINERY OCTANE BLEND MODELING USING PRINCIPAL COMPONENTS REGRESSION OF GAS-CHROMATOGRAPHY DATA

The use of capillary gas chromatography (g.c.) coupled with principal components regression (PCR) is shown to be an effective predictive tool for determining the octane values of gasolines blended from a wide variety of refinery streams. The linear combination of fuel stream g.c. data to give calculated blended fuel gas chromatograms and their associated octanes offers much promise for refinery blending optimization. © 1990.