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

A new theoretical model has been developed which explains the association b etween the molecular structure and the knock resistance of individual gasol ine compounds convincingly. The constitutions of more than 300 individual g asoline components were correlated with their knock rating (Blending Resear ch Octane Number, BRON) simultaneously. C-13 NMR spectra of all compounds w ere binned in 28 chemical shift regions of different size. The number of in dividual carbon signals of the nearly 2500 carbons was counted in each shif t region and was combined with the information about the presence or absenc e of the structure groups Oxygen, Rings, Aromatics, aliphatic Chains and ol efins (ORACL). These numbers were used for the encoding of the chemical str ucture. The relations between the structure information and the knock ratin gs were determined using an artificial neural network. For a validation dat a set of 50 individual chemical compounds from various substance classes co nsisting only of C, H and O a good agreement was found with their experimen tally determined BRON (R = 0.933). (C) 2001 Elsevier Science Ltd. All right s reserved.