Determination of octane numbers of gasoline compounds from their chemical structure by 13C NMR spectroscopy and neural networks

A new theoretical model has been developed which explains the association between the molecular structure and the knock resistance of individual gasoline compounds convincingly. The constitutions of more than 300 individual gasoline components were correlated with their knock rating (Blending Research Octane Number, BRON) simultaneously. 13C NMR spectra of all compounds were binned in 28 chemical shift regions of different size. The number of individual carbon signals of the nearly 2500 carbons was counted in each shift region and was combined with the information about the presence or absence of the structure groups Oxygen, Rings, Aromatics, aliphatic Chains and oLefins (ORACL). These numbers were used for the encoding of the chemical structure. The relations between the structure information and the knock ratings were determined using an artificial neural network. For a validation data set of 50 individual chemical compounds from various substance classes consisting only of C, H and O a good agreement was found with their experimentally determined BRON (R=0.933).