Answers are grounded in this paper and cite their sources.
Rapid prediction of gasoline quality parameters is essential for blend control and specification compliance; yet routine assessment still relies on conventional test methods that are time- and labour-intensive. In this study, Fourier-transform near infrared (FT-NIR) spectroscopy combined with chemometrics was evaluated using a targeted 5,240 to 6,280 cm −1 spectral interval centred on the C-H first overtone region for simultaneous prediction of five gasoline properties: density at 15°C, research octane number, Reid vapour pressure, percentage evaporated at 70°C, and ethyl tert -butyl ether (ETBE) as a targeted blend component. Principal component regression and partial least squares regression models were developed using multiplicative scatter correction, standard normal variate, and Savitzky-Golay second derivative (SD) pre-processing. Hold-out test results demonstrated strong predictive performance across investigated properties, with r 2 P ≥ 0.94, RMSEP values below reproducibility benchmarks of the corresponding reference methods, and RPD > 4 for most properties. Scatter and scale correction generally provided the best predictive performance, whereas SGSD showed no consistent advantage within the targeted interval. Notably, ETBE was quantified with high predictive performance (RMSEP ≈ 0.24 vol%; RPD ≈ 22), despite the smaller calibration dataset ( n = 73). Overall, the results indicate that the targeted C-H first-overtone interval contains sufficient spectral information for multi-property gasoline prediction within the investigated industrial dataset, supporting its potential use for rapid screening and refinery quality-control applications.
No takes yet. Share an insight, caveat, or question.
Othman et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: