Avaliação de correlações para propriedades físico-químicas de frações de petróleo

Abstract

Fuel fractions are complex mixtures consisting of a huge number of compounds. A complete description of all non-trace substances of these mixtures is experimentally unfeasible or computationally time consuming. For this reason, a complete molecular approach for calculation of properties of these mixtures is replaced by a pseudo-component approach or by correlations between bulk properties. Some of these properties are used in the regulation of fuel sells, e.g., for gasoline. Depending on the refining scheme and on the crude oils used to produce these fuels, a large variety of values are found for the properties of process streams that compose the final fuel. In order to accurately plan the blending of these streams, models have to be readily available for accurate calculation of the needed properties. In this work, eight sets of fuel Brazilians and two sets of fuel strange were analyzed: gasoline, kerosene, gas oil and diesel fractions. The properties analyzed for fractions were octane number, aromatics content, sulfur content, refractive index, density, pour point, flash point, freezing point, cloud point, aniline point, Reid vapor pressure and cetane number. Several correlations were evaluated, and the best performances were pointed out, allowing an accurate estimate of properties of the studied fuels. A parameter re-estimation procedure was applied, and new models were tuned against the experimental information. This technique allowed an accurate estimate of properties of the samples, verified by a considerable enhancement in the statistical performance of the models. Furthermore, the best model for each property and each series was presented