Estudo de alguns parâmetros físico-químicos de petróleos utilizando modelagem multivariada associada a espectroscopia no infravermelho

Abstract

Traditional methods of characterizing oils are laborious, expensive, require a long time to be realized and may result in environmental impacts. Studies have shown that the chemometrics associated with infrared spectroscopy has given very promising results in the characterizing parameters and validation of petroleum and its derivatives. Spectroscopic methods may be an effective alternative in the characterization of oil as a means of monitoring and determining physical and chemical parameters due to its potential, practical and analytical speed. In the first part of the methodology, spectroscopic measurements were performed in 99 oil samples using NIR and MIR and physical-chemical properties of oil were determined, such as total acid number (NAT), number of naphthenic acid (NAN), total nitrogen content (TNT) and basic nitrogen (TNB). In the second stage, the relevant information were extracted from the analytical methods employed by chemometric techniques such as PCA, PLS, iPLS and siPLS. The RMSEP of the model obtained for each property was 0.1964 mg KOH.g-1 for NAT, 0.3547 mg KOH.g-1 for NAN, 0.0381% m/m for TNT, 0.0073% m/m for TNB. The models for the acidity parameters can be improved, because they fail the trend tests on waste and permutation. The models for the properties related to nitrogen compounds obtained satisfactory results, as they passed all the validation criteria of the followed modeling protocol. The contribution of this work was the development of chemometric models by combining the information obtained by MIR and NIR spectroscopy with the results of oil characterization analysis, so that the models can anticipate the results by inference of the physicochemical characteristics of an oil, without the need to perform a complete characterization.