Ressonância magnética nuclear de baixo campo em estudos de petróleos

• Ano de defesa: 2009
• Autor(a) principal: Silva, Renzo Corrêa
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Química; Centro de Ciências Exatas; UFES; Programa de Pós-Graduação em Química
• Palavras-chave em Português: Water quantification; Petroleum blends; Chemometrics; Quantificação de água; Quimiometria; Misturas de petróleos; Ressonância magnética nuclear; Petróleo; Misturas (Química); Química; 54
• Link de acesso: http://repositorio.ufes.br/handle/10/4648

In some Nuclear Magnetic Resonance (NMR) applications the search for instruments capable of generating more intense magnetic fields is not necessary, turning the scientists to the development of low-field NMR equipments, typically of the benchtop kind: cheaper, lighter, with industrial appeal and suitable for on-line use. In the petroleum industry low-field NMR data are commonly used for in situ evaluations of petroleum formations (NMR logging), or to analyze rocks and fluids, on benchtop instruments at petrophysical laboratories. In this work, the potential of using benchtop low-field NMR instruments for analysis of mixtures involving crude oils and water was investigated. Experimental measurements were executed by means of the CPMG pulse sequence for T2 determination, and these data were used to obtain T2distribution curves by the Inverse Laplace Transformation. The instrument ability of quantifying water and petroleum in biphasic mixtures, by different methodologies, was tested. For deionized water and petroleum mixtures one achieved excellent results, with RMSECV (root mean squared error of cross-validation) of 0.8% for a regression between the water content (wt.%) and the relative area of the water peak in the T2 distribution curve, or a standard deviation of 0.9% for a relationship between the water content and the relative water peak area, corrected by the relative hydrogen index of the crude. In the case of biphasic mixtures of water doped with relaxing agent and crude oils, the best result of RMSECV=1.6% was achieved by using the raw magnetization decay data for a PLS regression. A related investigation involved the search for component quantification in crude oil blends; in this case, several mixtures of two distinct crude oils (with very different viscosities) were prepared. The quantification of each component in the mixture was accomplished with RMSECV=1.2% by using the T2 distribution curves for PLS regression. Three-component petroleum blends were also analyzed, being two of them very similar. A PLS model with 8 latent variables built from the T2 distribution data gave the RMSECV values of 4.1, 0.9 and 4.6% for the quantification of each oil component, showing the promising potential of the method for the study of more complex petroleum blends.