Spectroscopic investigations of crude oil and petroleum products in combination with chemometric methods

In the present work, chemometric methods were used for investigations of crude oils and petroleum products by statistical evaluation of spectroscopic data. The spectroscopic information was obtained by 1H-Nuclear Magnetic Resonance and infrared experiments. Research octane numbers (RON) and motor octane numbers (MON), representing important characteristics for knocking behavior,were calculated by multilinear regression. In the period of two years it was tested whether the precision of such models was negatively influenced by varying chemical composition of the commercial gasoline. A Partial Least Squares model (PLS) was created to calculate MON. The model based on infrared spectroscopic data in the spectral region of 1400 to 650 cm1. The standard deviation between PLS-method and the standard method ASTM D 2700 was 0,6. Another PLS-model in combination with spectroscopic data was used for calculating cetane numbers of diesel fuels. Using the spectral region of 1650 to 650 cm1 the content of the cetane improver 2-ethyl-hexyl-nitrate could be detected. A standard deviation of 1,8 to ASTM D 613 has been achieved. An additional aim of the work was to identify crude oils by the means of principal component analysis of spectroscopic data in the near infrared region. The investigations included the determination of negative effects on data acquisition like varying temperature, water-, salt-, and sediment content. Furthermore, the density and the boiling yields in distillation of crude oils were calculated with PLS-models. (author)