Fast assessment of diene values in pyrolysis gasoline by UV-visible spectroscopy and multivariate calibration

Conjugated dienes in pyrolysis gasoline samples were assayed by spectrophotometry. The calibrations were carried out by partial least squares regression using spectrophotometric data and the reference method UOP-326. The maximum deviation observed for an independent set of samples was 12% compared with the reference method. Spectral data preprocessing and weighting were investigated for the modeling. Values of root mean square error of calibration and prediction, and normalized root mean squared error were used as criteria to define the best calibration condition. 26 non-hydrogenated and 21 hydrogenated pyrolysis gasoline samples were used to model the spectral data by PLS regression.