Multivariate determination of several compositional parameters related to the content of hydrocarbon in naphtha by MIR spectroscopy

Abstract

Several compositional parameters of the main hydrocarbon families are determined in naphtha. This hydrocarbon mixture is traditionally analysed by gas chromatography (GC) but there are several alternative spectroscopic methods, such as near-infrared spectroscopy combined with multivariate calibration. Spectroscopic methods have the advantage that they are faster and more suitable for on-line analysis. Previous studies have determined the global percentage of each hydrocarbon family: linear and branched paraffins, naphthenes and aromatic compounds. Here we present the determination by mid-infrared (MIR) spectroscopy and multivariate calibration of the most detailed compositional parameters provided by gas chromatography. Results were good, that is to say there was no bias and the root mean square error of cross validation (RMSECV) was low, for the determination of all the detailed naphthenes and aromatic compounds, and most n-paraffins and isoparaffins. Several methods of detecting outliers are discussed and the quality of the models is evaluated by sample specific error predictions calculated by two approaches: the Unscrambler expression and the Faber–Kowalski expression. The proposed method enables a more specific analysis of naphtha than other methods and provides the same information as GC.